[{"data":1,"prerenderedAt":5366},["ShallowReactive",2],{"blog-\u002Fblog\u002F2026\u002F02\u002Fmotor-anomaly-detector-ai":3,"featureCatalog":4438,"changelog-titles":4831,"blog-all-for-related":5365},{"id":4,"title":5,"authors":6,"body":8,"cta":4413,"date":4417,"description":4418,"extension":4419,"features":4420,"image":4421,"lastUpdated":4422,"meta":4423,"navigation":655,"path":4428,"release":4420,"seo":4429,"sitemap":4430,"stem":4431,"subtitle":4432,"tags":4433,"tldr":4436,"video":4420,"__hash__":4437},"blog\u002Fblog\u002F2026\u002F02\u002Fmotor-anomaly-detector-ai.md","Building an AI Vibration Anomaly Detector for Industrial Motors",[7],"sumit-shinde",{"type":9,"value":10,"toc":4390},"minimark",[11,15,18,21,27,32,35,40,51,62,65,68,72,75,79,82,86,89,277,280,289,299,303,306,331,335,338,342,348,351,382,398,401,432,436,439,596,615,619,630,1879,1886,1907,1919,1923,1926,1930,1933,1937,1940,1958,1979,1986,1993,1997,2000,2054,2058,2061,2066,2073,2081,2086,2105,3539,3549,3560,3565,3570,3589,3595,3600,3606,4210,4221,4229,4232,4238,4243,4265,4269,4272,4347,4364,4368,4371,4374,4377,4383,4386],[12,13,14],"p",{},"Bearing wear, shaft misalignment, and imbalance don't appear overnight. They develop over days or weeks, leaving a clear trail in vibration data long before any audible or thermal symptoms emerge. By the time a technician hears grinding or feels heat, the window for low-cost intervention has already closed.",[12,16,17],{},"The challenge isn't visibility: it's continuity. Manual spot-checks capture a fraction of developing faults, and only if the timing is lucky. What's needed is something that watches constantly, understands what normal looks like, and flags the moment something shifts.",[12,19,20],{},"This guide walks through building exactly that: a custom AI model that learns the healthy vibration signature of your motor, detects deviations in real time, and integrates directly into Node-RED using FlowFuse with no separate ML infrastructure required.",[22,23],"lite-youtube",{"videoid":24,"style":25,"title":26},"Fkv2x3Kv0lY","width: 100%; aspect-ratio: 16\u002F9; background-image: url('\u002Fblog\u002F2026\u002F02\u002Fimages\u002Fanomaly-detection.png'); background-size: cover; background-position: center;","Motor Anomaly Detection System Built Using FlowFuse",[28,29,31],"h2",{"id":30},"how-it-works","How It Works",[12,33,34],{},"An accelerometer mounted on the motor captures vibration across three axes (X, Y, Z) and publishes batches of raw readings to an MQTT broker every half-second. A Node-RED flow subscribes to those readings, extracts 33 statistical features per batch (covering energy, peak forces, shape, and distribution across all three axes) and passes them to a trained autoencoder.",[36,37,39],"h3",{"id":38},"why-an-autoencoder","Why an Autoencoder?",[12,41,42,43,50],{},"An ",[44,45,49],"a",{"href":46,"rel":47},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FAutoencoder",[48],"nofollow","autoencoder"," is a neural network trained to compress its input and then reconstruct it. The architecture used here is:",[52,53,58],"pre",{"className":54,"code":56,"language":57},[55],"language-text","Input (33) → Dense (16) → Dense (8) → Dense (16) → Output (33)\n","text",[59,60,56],"code",{"__ignoreMap":61},"",[12,63,64],{},"The bottleneck layer (8 nodes) forces the model to learn a compact representation of the input. When trained exclusively on healthy motor data, the model learns to reconstruct normal vibration patterns with very low error. When conditions change (a bearing begins to wear, alignment drifts, imbalance develops) the vibration signature shifts, reconstruction error rises, and the system flags an anomaly.",[12,66,67],{},"This approach is well-suited to industrial use because you almost certainly have abundant examples of normal operation, but few or no labeled examples of specific fault modes. You don't need to know what failure looks like; you only need to define what normal looks like.",[28,69,71],{"id":70},"building-the-system","Building the System",[12,73,74],{},"The implementation has three stages: setting up hardware to collect vibration data, training the autoencoder on normal operation, and deploying the trained model in Node-RED for real-time inference.",[28,76,78],{"id":77},"part-1-hardware-and-data-requirements","Part 1: Hardware and Data Requirements",[12,80,81],{},"This guide assumes you already have a vibration sensor publishing batches of acceleration readings across X, Y, and Z axes at regular intervals. The examples were built using an ESP32 wired to an ADXL345 accelerometer. If your hardware differs, the rest of the steps remain unchanged as long as your sensor publishes the same payload format.",[36,83,85],{"id":84},"expected-payload-format","Expected Payload Format",[12,87,88],{},"Each MQTT message contains a half-second snapshot of motor vibration. The sensor captures 256 measurements per axis and packages them into a single JSON payload:",[52,90,94],{"className":91,"code":92,"language":93,"meta":61,"style":61},"language-json shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","{\n \"motor_id\": \"motor-01\",\n \"ts\": 1718000000000,\n \"x\": [0.12, 0.11, 0.13, 0.14, ...],\n \"y\": [0.04, 0.05, 0.04, 0.03, ...],\n \"z\": [0.98, 0.97, 0.99, 0.96, ...]\n}\n","json",[59,95,96,105,133,151,194,232,271],{"__ignoreMap":61},[97,98,101],"span",{"class":99,"line":100},"line",1,[97,102,104],{"class":103},"sMK4o","{\n",[97,106,108,111,115,118,121,124,128,130],{"class":99,"line":107},2,[97,109,110],{"class":103}," \"",[97,112,114],{"class":113},"spNyl","motor_id",[97,116,117],{"class":103},"\"",[97,119,120],{"class":103},":",[97,122,123],{"class":103}," \"",[97,125,127],{"class":126},"sfazB","motor-01",[97,129,117],{"class":103},[97,131,132],{"class":103},",\n",[97,134,136,138,141,143,145,149],{"class":99,"line":135},3,[97,137,110],{"class":103},[97,139,140],{"class":113},"ts",[97,142,117],{"class":103},[97,144,120],{"class":103},[97,146,148],{"class":147},"sbssI"," 1718000000000",[97,150,132],{"class":103},[97,152,154,156,159,161,163,166,169,172,175,177,180,182,185,187,191],{"class":99,"line":153},4,[97,155,110],{"class":103},[97,157,158],{"class":113},"x",[97,160,117],{"class":103},[97,162,120],{"class":103},[97,164,165],{"class":103}," [",[97,167,168],{"class":147},"0.12",[97,170,171],{"class":103},",",[97,173,174],{"class":147}," 0.11",[97,176,171],{"class":103},[97,178,179],{"class":147}," 0.13",[97,181,171],{"class":103},[97,183,184],{"class":147}," 0.14",[97,186,171],{"class":103},[97,188,190],{"class":189},"sTEyZ"," ...",[97,192,193],{"class":103},"],\n",[97,195,197,199,202,204,206,208,211,213,216,218,221,223,226,228,230],{"class":99,"line":196},5,[97,198,110],{"class":103},[97,200,201],{"class":113},"y",[97,203,117],{"class":103},[97,205,120],{"class":103},[97,207,165],{"class":103},[97,209,210],{"class":147},"0.04",[97,212,171],{"class":103},[97,214,215],{"class":147}," 0.05",[97,217,171],{"class":103},[97,219,220],{"class":147}," 0.04",[97,222,171],{"class":103},[97,224,225],{"class":147}," 0.03",[97,227,171],{"class":103},[97,229,190],{"class":189},[97,231,193],{"class":103},[97,233,235,237,240,242,244,246,249,251,254,256,259,261,264,266,268],{"class":99,"line":234},6,[97,236,110],{"class":103},[97,238,239],{"class":113},"z",[97,241,117],{"class":103},[97,243,120],{"class":103},[97,245,165],{"class":103},[97,247,248],{"class":147},"0.98",[97,250,171],{"class":103},[97,252,253],{"class":147}," 0.97",[97,255,171],{"class":103},[97,257,258],{"class":147}," 0.99",[97,260,171],{"class":103},[97,262,263],{"class":147}," 0.96",[97,265,171],{"class":103},[97,267,190],{"class":189},[97,269,270],{"class":103},"]\n",[97,272,274],{"class":99,"line":273},7,[97,275,276],{"class":103},"}\n",[12,278,279],{},"At 500 Hz sampling, 256 values represent roughly half a second of continuous vibration. This batching approach matters because it gives the AI model enough context to detect patterns: a single data point is meaningless, but 256 points reveal the behavioral signature of how the motor is actually running.",[12,281,282,283,285,286,288],{},"The ",[59,284,114],{}," and ",[59,287,140],{}," fields are ignored by the model and can be omitted or renamed without effect.",[290,291,292],"blockquote",{},[12,293,294,298],{},[295,296,297],"strong",{},"If your sensor uses different settings:"," The feature extraction math works regardless of sample count or sampling rate. If your sensor samples at 200 Hz and sends 128 values per batch, each window represents 640 ms instead of 500 ms; the model doesn't care about absolute timing, only the shape of the vibration signature. Aim for at least 100–200 ms of data per window; anything shorter may not carry enough signal for reliable detection.",[36,300,302],{"id":301},"mqtt-broker","MQTT Broker",[12,304,305],{},"You'll need an MQTT broker to route messages between the sensor, the training script, and Node-RED. Make sure your sensor is publishing to a consistent topic so all three can stay in sync.",[290,307,308],{},[12,309,310,313,314,318,319,285,322,325,326,330],{},[295,311,312],{},"Tip:"," If you're using ",[44,315,317],{"href":316},"\u002F","FlowFuse"," for enterprise Node-RED, a built-in MQTT broker is available on ",[295,320,321],{},"Pro",[295,323,324],{},"Enterprise"," tiers with no external setup required. ",[44,327,329],{"href":328},"\u002Fcontact-us","Contact us"," for more information.",[28,332,334],{"id":333},"part-2-training-the-autoencoder","Part 2: Training the Autoencoder",[12,336,337],{},"Before deploying anything in Node-RED, you need a trained model that understands what normal motor vibration looks like. This is done with a single Python script that connects to your MQTT broker, collects vibration data while the motor runs normally, then automatically trains and exports the model when you're done.",[36,339,341],{"id":340},"prerequisites","Prerequisites",[12,343,344,347],{},[295,345,346],{},"System requirements:"," Python 3.11 or later. The steps below were tested on macOS (Apple Silicon); adapt as needed for Linux or Windows.",[12,349,350],{},"Create and activate a virtual environment:",[52,352,356],{"className":353,"code":354,"language":355,"meta":61,"style":61},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","python3 -m venv venv\nsource venv\u002Fbin\u002Factivate\n","bash",[59,357,358,373],{"__ignoreMap":61},[97,359,360,364,367,370],{"class":99,"line":100},[97,361,363],{"class":362},"sBMFI","python3",[97,365,366],{"class":126}," -m",[97,368,369],{"class":126}," venv",[97,371,372],{"class":126}," venv\n",[97,374,375,379],{"class":99,"line":107},[97,376,378],{"class":377},"s2Zo4","source",[97,380,381],{"class":126}," venv\u002Fbin\u002Factivate\n",[290,383,384],{},[12,385,386,389,390,393,394,397],{},[295,387,388],{},"Windows:"," Replace ",[59,391,392],{},"source venv\u002Fbin\u002Factivate"," with ",[59,395,396],{},"venv\\Scripts\\activate",".",[12,399,400],{},"Install dependencies:",[52,402,404],{"className":353,"code":403,"language":355,"meta":61,"style":61},"pip3 install numpy paho-mqtt torch onnx onnxruntime scikit-learn\n",[59,405,406],{"__ignoreMap":61},[97,407,408,411,414,417,420,423,426,429],{"class":99,"line":100},[97,409,410],{"class":362},"pip3",[97,412,413],{"class":126}," install",[97,415,416],{"class":126}," numpy",[97,418,419],{"class":126}," paho-mqtt",[97,421,422],{"class":126}," torch",[97,424,425],{"class":126}," onnx",[97,427,428],{"class":126}," onnxruntime",[97,430,431],{"class":126}," scikit-learn\n",[36,433,435],{"id":434},"configuration","Configuration",[12,437,438],{},"All setup lives in a single configuration block at the top of the script. Before running, update these variables to match your environment:",[440,441,442,455],"table",{},[443,444,445],"thead",{},[446,447,448,452],"tr",{},[449,450,451],"th",{},"Variable",[449,453,454],{},"Description",[456,457,458,469,486,500,512,522,532,542,552,562,572,582],"tbody",{},[446,459,460,466],{},[461,462,463],"td",{},[59,464,465],{},"BROKER",[461,467,468],{},"Hostname or IP of your MQTT broker",[446,470,471,476],{},[461,472,473],{},[59,474,475],{},"PORT",[461,477,478,481,482,485],{},[59,479,480],{},"1883"," for plain MQTT, ",[59,483,484],{},"8883"," for TLS",[446,487,488,493],{},[461,489,490],{},[59,491,492],{},"USERNAME",[461,494,495,496,499],{},"Broker username. Leave empty ",[59,497,498],{},"\"\""," if not required",[446,501,502,507],{},[461,503,504],{},[59,505,506],{},"PASSWORD",[461,508,509,510,499],{},"Broker password. Leave empty ",[59,511,498],{},[446,513,514,519],{},[461,515,516],{},[59,517,518],{},"CLIENT_ID",[461,520,521],{},"Any unique string identifying this client",[446,523,524,529],{},[461,525,526],{},[59,527,528],{},"TOPIC",[461,530,531],{},"The MQTT topic your sensor publishes to",[446,533,534,539],{},[461,535,536],{},[59,537,538],{},"MIN_WINDOWS",[461,540,541],{},"Minimum samples to collect before training (default: 300)",[446,543,544,549],{},[461,545,546],{},[59,547,548],{},"MIN_STD",[461,550,551],{},"Minimum standard deviation floor that prevents near-constant features from skewing normalisation (default: 0.1)",[446,553,554,559],{},[461,555,556],{},[59,557,558],{},"CLIP",[461,560,561],{},"Hard clamp applied after normalisation to prevent extreme values (default: 5.0)",[446,563,564,569],{},[461,565,566],{},[59,567,568],{},"EPOCHS",[461,570,571],{},"Number of training epochs (default: 200)",[446,573,574,579],{},[461,575,576],{},[59,577,578],{},"LEARNING_RATE",[461,580,581],{},"Adam optimizer learning rate (default: 0.001)",[446,583,584,589],{},[461,585,586],{},[59,587,588],{},"THRESHOLD_SIGMA",[461,590,591,592,595],{},"Multiplier for threshold calculation: ",[59,593,594],{},"mean + N × std"," of training errors (default: 3)",[290,597,598],{},[12,599,600,603,604,607,608,610,611,614],{},[295,601,602],{},"Threshold tuning:"," The default ",[59,605,606],{},"mean + 3σ"," threshold is a solid starting point, but every motor environment is different. If you see too many false positives during normal operation, increase ",[59,609,588],{},". If faults are being missed, decrease it. You can also edit ",[59,612,613],{},"threshold.json"," directly after training without rerunning the script.",[36,616,618],{"id":617},"collect-and-train","Collect and Train",[12,620,621,622,625,626,629],{},"Create a file called ",[59,623,624],{},"train_model.py"," and paste the following. ",[295,627,628],{},"Start the motor first, then run the script."," The model needs to learn what running vibration looks like. Collecting data with the motor stopped or barely loaded will produce a model that treats idle conditions as normal and misses real anomalies.",[52,631,635],{"className":632,"code":633,"language":634,"meta":61,"style":61},"language-python shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","\"\"\"\nMotor Vibration Anomaly Detection: Data Collection and Training\n\"\"\"\n\nimport json\nimport signal\nimport sys\nimport numpy as np\nimport torch\nimport torch.nn as nn\nimport onnx\nimport onnxruntime as ort\nfrom onnx import numpy_helper, TensorProto, helper\nimport paho.mqtt.client as mqtt\nfrom paho.mqtt.client import CallbackAPIVersion\n\n# ── Configuration ────────────────────────────────────────────────────────────\nBROKER = \"broker.example.com\" # Your MQTT broker address or IP\nPORT = 1883 # 1883 for plain MQTT, 8883 for TLS\nUSERNAME = \"\" # Leave as \"\" if broker has no auth\nPASSWORD = \"\" # Leave as \"\" if broker has no auth\nCLIENT_ID = \"motor-trainer-01\" # Any unique string for this client\nTOPIC = \"factory\u002Fmotor-01\u002Fvibration\u002Fraw\" # Must match your sensor's publish topic\n\n# ── Training parameters ───────────────────────────────────────────────────────\nMIN_WINDOWS = 300 # Minimum samples before training is allowed\nMIN_STD = 0.1 # Prevents near-constant features from exploding normalisation\nCLIP = 5.0 # Hard clip applied after normalisation\nEPOCHS = 200 # Number of training epochs\nLEARNING_RATE = 1e-3 # Adam optimiser learning rate\nTHRESHOLD_SIGMA = 3 # Threshold = mean + N * std of training reconstruction errors\n# ─────────────────────────────────────────────────────────────────────────────\n\ntraining_data = []\nstop_flag = [False]\n\ndef extract_features(sig):\n \"\"\"Extract 11 time-domain features from a signal array.\"\"\"\n sig = np.asarray(sig, dtype=np.float64)\n mean = np.mean(sig)\n std = np.std(sig) + 1e-9\n\n rms = np.sqrt(np.mean(sig ** 2))\n peak = np.max(np.abs(sig))\n peak_to_peak = np.max(sig) - np.min(sig)\n crest_factor = peak \u002F (rms + 1e-9)\n variance = np.var(sig)\n std_dev = np.std(sig)\n skewness = np.mean(((sig - mean) \u002F std) ** 3)\n kurtosis = np.mean(((sig - mean) \u002F std) ** 4) - 3\n mean_abs = np.mean(np.abs(sig)) + 1e-9\n shape_factor = rms \u002F mean_abs\n impulse_factor = peak \u002F mean_abs\n mean_sqrt_abs = np.mean(np.sqrt(np.abs(sig)))\n clearance_factor = peak \u002F (mean_sqrt_abs ** 2 + 1e-9)\n\n return [rms, peak, peak_to_peak, crest_factor, variance,\n std_dev, skewness, kurtosis, shape_factor,\n impulse_factor, clearance_factor]\n\ndef featurize(payload):\n \"\"\"Concatenate features from X, Y, Z → 33-element vector.\"\"\"\n return (extract_features(payload[\"x\"]) +\n extract_features(payload[\"y\"]) +\n extract_features(payload[\"z\"]))\n\ndef on_connect(client, userdata, flags, reason_code, properties):\n if reason_code == 0:\n print(f\"Connected to {BROKER}\")\n client.subscribe(TOPIC)\n print(f\"Subscribed to {TOPIC}\")\n print(\"Run motor normally. Press Ctrl+C when done collecting.\\n\")\n else:\n print(f\"Connection failed: {reason_code}\")\n\ndef on_message(client, userdata, msg):\n if stop_flag[0]:\n return\n try:\n payload = json.loads(msg.payload.decode())\n training_data.append(featurize(payload))\n n = len(training_data)\n print(f\" Collected {n} windows\", end=\"\\r\")\n except Exception as e:\n print(f\"\\nError parsing message: {e}\")\n\ndef train_and_export():\n N = 33\n print(f\"\\n\\nCollected {len(training_data)} windows. Starting training...\")\n X = np.array(training_data, dtype=np.float32)\n\n # Normalisation with minimum std floor\n mean = X.mean(axis=0)\n std = X.std(axis=0)\n\n clamped = std \u003C MIN_STD\n if clamped.any():\n print(f\" Clamping {clamped.sum()} near-constant features to std={MIN_STD}\")\n std[clamped] = MIN_STD\n\n X_norm = np.clip((X - mean) \u002F std, -CLIP, CLIP)\n print(f\" Normalised range: [{X_norm.min():.3f}, {X_norm.max():.3f}]\")\n\n scaler = {\"mean\": mean.tolist(), \"std\": std.tolist(), \"clip\": CLIP}\n with open(\"scaler_params.json\", \"w\") as f:\n json.dump(scaler, f, indent=2)\n print(\" Saved scaler_params.json\")\n\n # Autoencoder definition\n class Autoencoder(nn.Module):\n def __init__(self, n):\n super().__init__()\n self.encoder = nn.Sequential(\n nn.Linear(n, 16), nn.ReLU(),\n nn.Linear(16, 8), nn.ReLU(),\n )\n self.decoder = nn.Sequential(\n nn.Linear(8, 16), nn.ReLU(),\n nn.Linear(16, n),\n )\n def forward(self, x):\n return self.decoder(self.encoder(x))\n\n model = Autoencoder(N)\n opt = torch.optim.Adam(model.parameters(), lr=LEARNING_RATE)\n loss_fn = nn.MSELoss()\n data_t = torch.tensor(X_norm, dtype=torch.float32)\n\n model.train()\n for epoch in range(1, EPOCHS + 1):\n opt.zero_grad()\n loss = loss_fn(model(data_t), data_t)\n loss.backward()\n opt.step()\n if epoch % (EPOCHS \u002F\u002F 5) == 0:\n print(f\" Epoch {epoch}\u002F{EPOCHS} loss={loss.item():.6f}\")\n\n # Calculate threshold\n model.eval()\n with torch.no_grad():\n recon = model(data_t).numpy()\n errors = np.mean((recon - X_norm) ** 2, axis=1)\n thresh = float(errors.mean() + THRESHOLD_SIGMA * errors.std())\n\n with open(\"threshold.json\", \"w\") as f:\n json.dump({\"threshold\": thresh}, f, indent=2)\n print(f\" Threshold (mean+{THRESHOLD_SIGMA}σ): {thresh:.6f}\")\n print(\" Saved threshold.json\")\n\n # Export to ONNX\n # Built manually to avoid version conflicts between PyTorch and ONNX exporters\n layers = [(\"encoder.0\",\"enc0\"),(\"encoder.2\",\"enc2\"),\n (\"decoder.0\",\"dec0\"),(\"decoder.2\",\"dec2\")]\n has_relu = [True, True, True, False]\n inits, nodes = [], []\n cur = \"features\"\n\n for (prefix, tag), relu in zip(layers, has_relu):\n w = model.state_dict()[f\"{prefix}.weight\"].numpy().T.astype(np.float32)\n b = model.state_dict()[f\"{prefix}.bias\"].numpy().astype(np.float32)\n inits += [numpy_helper.from_array(w, name=f\"w_{tag}\"),\n numpy_helper.from_array(b, name=f\"b_{tag}\")]\n mm = f\"mm_{tag}\"; add = f\"add_{tag}\"\n nodes += [helper.make_node(\"MatMul\", [cur, f\"w_{tag}\"], [mm]),\n helper.make_node(\"Add\", [mm, f\"b_{tag}\"], [add])]\n cur = add\n if relu:\n r = f\"relu_{tag}\"\n nodes.append(helper.make_node(\"Relu\", [cur], [r]))\n cur = r\n\n graph = helper.make_graph(\n nodes, \"autoencoder\",\n [helper.make_tensor_value_info(\"features\", TensorProto.FLOAT, [None, N])],\n [helper.make_tensor_value_info(cur, TensorProto.FLOAT, [None, N])],\n initializer=inits,\n )\n proto = helper.make_model(graph, opset_imports=[helper.make_opsetid(\"\", 11)])\n proto.ir_version = 7\n onnx.checker.check_model(proto)\n onnx.save(proto, \"motor_autoencoder.onnx\")\n print(\" Exported motor_autoencoder.onnx\")\n\n # Sanity check\n sess = ort.InferenceSession(\"motor_autoencoder.onnx\")\n out = sess.run(None, {\"features\": X_norm[:5].astype(np.float32)})[0]\n mse = float(np.mean((out - X_norm[:5]) ** 2))\n print(f\"\\n Sanity MSE (5 normal samples): {mse:.6f} threshold: {thresh:.6f}\")\n if mse \u003C thresh:\n print(\" ✓ Model correct , normal data scores below threshold.\")\n else:\n print(\" ⚠ Sanity MSE above threshold , collect more data and retrain.\")\n\n print(\"\\nDone. Copy these 3 files to your Node-RED server:\")\n print(\" motor_autoencoder.onnx scaler_params.json threshold.json\")\n\ndef handle_sigint(sig, frame):\n stop_flag[0] = True\n if len(training_data) \u003C MIN_WINDOWS:\n print(f\"\\n\\nNeed at least {MIN_WINDOWS} windows. Restart and collect longer.\")\n sys.exit(1)\n train_and_export()\n sys.exit(0)\n\nsignal.signal(signal.SIGINT, handle_sigint)\n\nclient = mqtt.Client(callback_api_version=CallbackAPIVersion.VERSION2,\n client_id=CLIENT_ID)\nclient.username_pw_set(USERNAME, PASSWORD)\nclient.on_connect = on_connect\nclient.on_message = on_message\nclient.connect(BROKER, PORT, keepalive=60)\nclient.loop_forever()\n","python",[59,636,637,642,647,651,657,662,667,672,678,684,690,696,702,708,714,720,725,731,737,743,749,755,761,767,772,778,784,790,796,802,808,814,820,825,831,837,842,848,854,860,866,872,877,883,889,895,901,907,913,919,925,931,937,943,949,955,960,966,972,978,983,989,995,1001,1007,1013,1018,1024,1030,1036,1042,1048,1054,1060,1066,1071,1077,1083,1089,1095,1101,1107,1113,1119,1125,1131,1136,1142,1148,1154,1160,1165,1171,1177,1183,1188,1194,1200,1206,1212,1217,1223,1229,1234,1240,1246,1252,1258,1263,1269,1275,1281,1287,1293,1299,1305,1311,1317,1323,1329,1334,1340,1346,1351,1357,1363,1369,1375,1380,1386,1392,1398,1404,1410,1416,1422,1428,1433,1439,1445,1451,1457,1463,1469,1474,1480,1486,1492,1498,1503,1509,1515,1521,1527,1533,1539,1545,1550,1556,1562,1568,1574,1580,1586,1592,1598,1604,1610,1616,1622,1628,1633,1639,1645,1651,1657,1663,1669,1675,1681,1687,1693,1699,1704,1710,1716,1722,1728,1734,1740,1746,1751,1757,1762,1768,1774,1779,1785,1791,1797,1803,1809,1815,1821,1826,1832,1837,1843,1849,1855,1861,1867,1873],{"__ignoreMap":61},[97,638,639],{"class":99,"line":100},[97,640,641],{},"\"\"\"\n",[97,643,644],{"class":99,"line":107},[97,645,646],{},"Motor Vibration Anomaly Detection: Data Collection and Training\n",[97,648,649],{"class":99,"line":135},[97,650,641],{},[97,652,653],{"class":99,"line":153},[97,654,656],{"emptyLinePlaceholder":655},true,"\n",[97,658,659],{"class":99,"line":196},[97,660,661],{},"import json\n",[97,663,664],{"class":99,"line":234},[97,665,666],{},"import signal\n",[97,668,669],{"class":99,"line":273},[97,670,671],{},"import sys\n",[97,673,675],{"class":99,"line":674},8,[97,676,677],{},"import numpy as np\n",[97,679,681],{"class":99,"line":680},9,[97,682,683],{},"import torch\n",[97,685,687],{"class":99,"line":686},10,[97,688,689],{},"import torch.nn as nn\n",[97,691,693],{"class":99,"line":692},11,[97,694,695],{},"import onnx\n",[97,697,699],{"class":99,"line":698},12,[97,700,701],{},"import onnxruntime as ort\n",[97,703,705],{"class":99,"line":704},13,[97,706,707],{},"from onnx import numpy_helper, TensorProto, helper\n",[97,709,711],{"class":99,"line":710},14,[97,712,713],{},"import paho.mqtt.client as mqtt\n",[97,715,717],{"class":99,"line":716},15,[97,718,719],{},"from paho.mqtt.client import CallbackAPIVersion\n",[97,721,723],{"class":99,"line":722},16,[97,724,656],{"emptyLinePlaceholder":655},[97,726,728],{"class":99,"line":727},17,[97,729,730],{},"# ── Configuration ────────────────────────────────────────────────────────────\n",[97,732,734],{"class":99,"line":733},18,[97,735,736],{},"BROKER = \"broker.example.com\" # Your MQTT broker address or IP\n",[97,738,740],{"class":99,"line":739},19,[97,741,742],{},"PORT = 1883 # 1883 for plain MQTT, 8883 for TLS\n",[97,744,746],{"class":99,"line":745},20,[97,747,748],{},"USERNAME = \"\" # Leave as \"\" if broker has no auth\n",[97,750,752],{"class":99,"line":751},21,[97,753,754],{},"PASSWORD = \"\" # Leave as \"\" if broker has no auth\n",[97,756,758],{"class":99,"line":757},22,[97,759,760],{},"CLIENT_ID = \"motor-trainer-01\" # Any unique string for this client\n",[97,762,764],{"class":99,"line":763},23,[97,765,766],{},"TOPIC = \"factory\u002Fmotor-01\u002Fvibration\u002Fraw\" # Must match your sensor's publish topic\n",[97,768,770],{"class":99,"line":769},24,[97,771,656],{"emptyLinePlaceholder":655},[97,773,775],{"class":99,"line":774},25,[97,776,777],{},"# ── Training parameters ───────────────────────────────────────────────────────\n",[97,779,781],{"class":99,"line":780},26,[97,782,783],{},"MIN_WINDOWS = 300 # Minimum samples before training is allowed\n",[97,785,787],{"class":99,"line":786},27,[97,788,789],{},"MIN_STD = 0.1 # Prevents near-constant features from exploding normalisation\n",[97,791,793],{"class":99,"line":792},28,[97,794,795],{},"CLIP = 5.0 # Hard clip applied after normalisation\n",[97,797,799],{"class":99,"line":798},29,[97,800,801],{},"EPOCHS = 200 # Number of training epochs\n",[97,803,805],{"class":99,"line":804},30,[97,806,807],{},"LEARNING_RATE = 1e-3 # Adam optimiser learning rate\n",[97,809,811],{"class":99,"line":810},31,[97,812,813],{},"THRESHOLD_SIGMA = 3 # Threshold = mean + N * std of training reconstruction errors\n",[97,815,817],{"class":99,"line":816},32,[97,818,819],{},"# ─────────────────────────────────────────────────────────────────────────────\n",[97,821,823],{"class":99,"line":822},33,[97,824,656],{"emptyLinePlaceholder":655},[97,826,828],{"class":99,"line":827},34,[97,829,830],{},"training_data = []\n",[97,832,834],{"class":99,"line":833},35,[97,835,836],{},"stop_flag = [False]\n",[97,838,840],{"class":99,"line":839},36,[97,841,656],{"emptyLinePlaceholder":655},[97,843,845],{"class":99,"line":844},37,[97,846,847],{},"def extract_features(sig):\n",[97,849,851],{"class":99,"line":850},38,[97,852,853],{}," \"\"\"Extract 11 time-domain features from a signal array.\"\"\"\n",[97,855,857],{"class":99,"line":856},39,[97,858,859],{}," sig = np.asarray(sig, dtype=np.float64)\n",[97,861,863],{"class":99,"line":862},40,[97,864,865],{}," mean = np.mean(sig)\n",[97,867,869],{"class":99,"line":868},41,[97,870,871],{}," std = np.std(sig) + 1e-9\n",[97,873,875],{"class":99,"line":874},42,[97,876,656],{"emptyLinePlaceholder":655},[97,878,880],{"class":99,"line":879},43,[97,881,882],{}," rms = np.sqrt(np.mean(sig ** 2))\n",[97,884,886],{"class":99,"line":885},44,[97,887,888],{}," peak = np.max(np.abs(sig))\n",[97,890,892],{"class":99,"line":891},45,[97,893,894],{}," peak_to_peak = np.max(sig) - np.min(sig)\n",[97,896,898],{"class":99,"line":897},46,[97,899,900],{}," crest_factor = peak \u002F (rms + 1e-9)\n",[97,902,904],{"class":99,"line":903},47,[97,905,906],{}," variance = np.var(sig)\n",[97,908,910],{"class":99,"line":909},48,[97,911,912],{}," std_dev = np.std(sig)\n",[97,914,916],{"class":99,"line":915},49,[97,917,918],{}," skewness = np.mean(((sig - mean) \u002F std) ** 3)\n",[97,920,922],{"class":99,"line":921},50,[97,923,924],{}," kurtosis = np.mean(((sig - mean) \u002F std) ** 4) - 3\n",[97,926,928],{"class":99,"line":927},51,[97,929,930],{}," mean_abs = np.mean(np.abs(sig)) + 1e-9\n",[97,932,934],{"class":99,"line":933},52,[97,935,936],{}," shape_factor = rms \u002F mean_abs\n",[97,938,940],{"class":99,"line":939},53,[97,941,942],{}," impulse_factor = peak \u002F mean_abs\n",[97,944,946],{"class":99,"line":945},54,[97,947,948],{}," mean_sqrt_abs = np.mean(np.sqrt(np.abs(sig)))\n",[97,950,952],{"class":99,"line":951},55,[97,953,954],{}," clearance_factor = peak \u002F (mean_sqrt_abs ** 2 + 1e-9)\n",[97,956,958],{"class":99,"line":957},56,[97,959,656],{"emptyLinePlaceholder":655},[97,961,963],{"class":99,"line":962},57,[97,964,965],{}," return [rms, peak, peak_to_peak, crest_factor, variance,\n",[97,967,969],{"class":99,"line":968},58,[97,970,971],{}," std_dev, skewness, kurtosis, shape_factor,\n",[97,973,975],{"class":99,"line":974},59,[97,976,977],{}," impulse_factor, clearance_factor]\n",[97,979,981],{"class":99,"line":980},60,[97,982,656],{"emptyLinePlaceholder":655},[97,984,986],{"class":99,"line":985},61,[97,987,988],{},"def featurize(payload):\n",[97,990,992],{"class":99,"line":991},62,[97,993,994],{}," \"\"\"Concatenate features from X, Y, Z → 33-element vector.\"\"\"\n",[97,996,998],{"class":99,"line":997},63,[97,999,1000],{}," return (extract_features(payload[\"x\"]) +\n",[97,1002,1004],{"class":99,"line":1003},64,[97,1005,1006],{}," extract_features(payload[\"y\"]) +\n",[97,1008,1010],{"class":99,"line":1009},65,[97,1011,1012],{}," extract_features(payload[\"z\"]))\n",[97,1014,1016],{"class":99,"line":1015},66,[97,1017,656],{"emptyLinePlaceholder":655},[97,1019,1021],{"class":99,"line":1020},67,[97,1022,1023],{},"def on_connect(client, userdata, flags, reason_code, properties):\n",[97,1025,1027],{"class":99,"line":1026},68,[97,1028,1029],{}," if reason_code == 0:\n",[97,1031,1033],{"class":99,"line":1032},69,[97,1034,1035],{}," print(f\"Connected to {BROKER}\")\n",[97,1037,1039],{"class":99,"line":1038},70,[97,1040,1041],{}," client.subscribe(TOPIC)\n",[97,1043,1045],{"class":99,"line":1044},71,[97,1046,1047],{}," print(f\"Subscribed to {TOPIC}\")\n",[97,1049,1051],{"class":99,"line":1050},72,[97,1052,1053],{}," print(\"Run motor normally. Press Ctrl+C when done collecting.\\n\")\n",[97,1055,1057],{"class":99,"line":1056},73,[97,1058,1059],{}," else:\n",[97,1061,1063],{"class":99,"line":1062},74,[97,1064,1065],{}," print(f\"Connection failed: {reason_code}\")\n",[97,1067,1069],{"class":99,"line":1068},75,[97,1070,656],{"emptyLinePlaceholder":655},[97,1072,1074],{"class":99,"line":1073},76,[97,1075,1076],{},"def on_message(client, userdata, msg):\n",[97,1078,1080],{"class":99,"line":1079},77,[97,1081,1082],{}," if stop_flag[0]:\n",[97,1084,1086],{"class":99,"line":1085},78,[97,1087,1088],{}," return\n",[97,1090,1092],{"class":99,"line":1091},79,[97,1093,1094],{}," try:\n",[97,1096,1098],{"class":99,"line":1097},80,[97,1099,1100],{}," payload = json.loads(msg.payload.decode())\n",[97,1102,1104],{"class":99,"line":1103},81,[97,1105,1106],{}," training_data.append(featurize(payload))\n",[97,1108,1110],{"class":99,"line":1109},82,[97,1111,1112],{}," n = len(training_data)\n",[97,1114,1116],{"class":99,"line":1115},83,[97,1117,1118],{}," print(f\" Collected {n} windows\", end=\"\\r\")\n",[97,1120,1122],{"class":99,"line":1121},84,[97,1123,1124],{}," except Exception as e:\n",[97,1126,1128],{"class":99,"line":1127},85,[97,1129,1130],{}," print(f\"\\nError parsing message: {e}\")\n",[97,1132,1134],{"class":99,"line":1133},86,[97,1135,656],{"emptyLinePlaceholder":655},[97,1137,1139],{"class":99,"line":1138},87,[97,1140,1141],{},"def train_and_export():\n",[97,1143,1145],{"class":99,"line":1144},88,[97,1146,1147],{}," N = 33\n",[97,1149,1151],{"class":99,"line":1150},89,[97,1152,1153],{}," print(f\"\\n\\nCollected {len(training_data)} windows. Starting training...\")\n",[97,1155,1157],{"class":99,"line":1156},90,[97,1158,1159],{}," X = np.array(training_data, dtype=np.float32)\n",[97,1161,1163],{"class":99,"line":1162},91,[97,1164,656],{"emptyLinePlaceholder":655},[97,1166,1168],{"class":99,"line":1167},92,[97,1169,1170],{}," # Normalisation with minimum std floor\n",[97,1172,1174],{"class":99,"line":1173},93,[97,1175,1176],{}," mean = X.mean(axis=0)\n",[97,1178,1180],{"class":99,"line":1179},94,[97,1181,1182],{}," std = X.std(axis=0)\n",[97,1184,1186],{"class":99,"line":1185},95,[97,1187,656],{"emptyLinePlaceholder":655},[97,1189,1191],{"class":99,"line":1190},96,[97,1192,1193],{}," clamped = std \u003C MIN_STD\n",[97,1195,1197],{"class":99,"line":1196},97,[97,1198,1199],{}," if clamped.any():\n",[97,1201,1203],{"class":99,"line":1202},98,[97,1204,1205],{}," print(f\" Clamping {clamped.sum()} near-constant features to std={MIN_STD}\")\n",[97,1207,1209],{"class":99,"line":1208},99,[97,1210,1211],{}," std[clamped] = MIN_STD\n",[97,1213,1215],{"class":99,"line":1214},100,[97,1216,656],{"emptyLinePlaceholder":655},[97,1218,1220],{"class":99,"line":1219},101,[97,1221,1222],{}," X_norm = np.clip((X - mean) \u002F std, -CLIP, CLIP)\n",[97,1224,1226],{"class":99,"line":1225},102,[97,1227,1228],{}," print(f\" Normalised range: [{X_norm.min():.3f}, {X_norm.max():.3f}]\")\n",[97,1230,1232],{"class":99,"line":1231},103,[97,1233,656],{"emptyLinePlaceholder":655},[97,1235,1237],{"class":99,"line":1236},104,[97,1238,1239],{}," scaler = {\"mean\": mean.tolist(), \"std\": std.tolist(), \"clip\": CLIP}\n",[97,1241,1243],{"class":99,"line":1242},105,[97,1244,1245],{}," with open(\"scaler_params.json\", \"w\") as f:\n",[97,1247,1249],{"class":99,"line":1248},106,[97,1250,1251],{}," json.dump(scaler, f, indent=2)\n",[97,1253,1255],{"class":99,"line":1254},107,[97,1256,1257],{}," print(\" Saved scaler_params.json\")\n",[97,1259,1261],{"class":99,"line":1260},108,[97,1262,656],{"emptyLinePlaceholder":655},[97,1264,1266],{"class":99,"line":1265},109,[97,1267,1268],{}," # Autoencoder definition\n",[97,1270,1272],{"class":99,"line":1271},110,[97,1273,1274],{}," class Autoencoder(nn.Module):\n",[97,1276,1278],{"class":99,"line":1277},111,[97,1279,1280],{}," def __init__(self, n):\n",[97,1282,1284],{"class":99,"line":1283},112,[97,1285,1286],{}," super().__init__()\n",[97,1288,1290],{"class":99,"line":1289},113,[97,1291,1292],{}," self.encoder = nn.Sequential(\n",[97,1294,1296],{"class":99,"line":1295},114,[97,1297,1298],{}," nn.Linear(n, 16), nn.ReLU(),\n",[97,1300,1302],{"class":99,"line":1301},115,[97,1303,1304],{}," nn.Linear(16, 8), nn.ReLU(),\n",[97,1306,1308],{"class":99,"line":1307},116,[97,1309,1310],{}," )\n",[97,1312,1314],{"class":99,"line":1313},117,[97,1315,1316],{}," self.decoder = nn.Sequential(\n",[97,1318,1320],{"class":99,"line":1319},118,[97,1321,1322],{}," nn.Linear(8, 16), nn.ReLU(),\n",[97,1324,1326],{"class":99,"line":1325},119,[97,1327,1328],{}," nn.Linear(16, n),\n",[97,1330,1332],{"class":99,"line":1331},120,[97,1333,1310],{},[97,1335,1337],{"class":99,"line":1336},121,[97,1338,1339],{}," def forward(self, x):\n",[97,1341,1343],{"class":99,"line":1342},122,[97,1344,1345],{}," return self.decoder(self.encoder(x))\n",[97,1347,1349],{"class":99,"line":1348},123,[97,1350,656],{"emptyLinePlaceholder":655},[97,1352,1354],{"class":99,"line":1353},124,[97,1355,1356],{}," model = Autoencoder(N)\n",[97,1358,1360],{"class":99,"line":1359},125,[97,1361,1362],{}," opt = torch.optim.Adam(model.parameters(), lr=LEARNING_RATE)\n",[97,1364,1366],{"class":99,"line":1365},126,[97,1367,1368],{}," loss_fn = nn.MSELoss()\n",[97,1370,1372],{"class":99,"line":1371},127,[97,1373,1374],{}," data_t = torch.tensor(X_norm, dtype=torch.float32)\n",[97,1376,1378],{"class":99,"line":1377},128,[97,1379,656],{"emptyLinePlaceholder":655},[97,1381,1383],{"class":99,"line":1382},129,[97,1384,1385],{}," model.train()\n",[97,1387,1389],{"class":99,"line":1388},130,[97,1390,1391],{}," for epoch in range(1, EPOCHS + 1):\n",[97,1393,1395],{"class":99,"line":1394},131,[97,1396,1397],{}," opt.zero_grad()\n",[97,1399,1401],{"class":99,"line":1400},132,[97,1402,1403],{}," loss = loss_fn(model(data_t), data_t)\n",[97,1405,1407],{"class":99,"line":1406},133,[97,1408,1409],{}," loss.backward()\n",[97,1411,1413],{"class":99,"line":1412},134,[97,1414,1415],{}," opt.step()\n",[97,1417,1419],{"class":99,"line":1418},135,[97,1420,1421],{}," if epoch % (EPOCHS \u002F\u002F 5) == 0:\n",[97,1423,1425],{"class":99,"line":1424},136,[97,1426,1427],{}," print(f\" Epoch {epoch}\u002F{EPOCHS} loss={loss.item():.6f}\")\n",[97,1429,1431],{"class":99,"line":1430},137,[97,1432,656],{"emptyLinePlaceholder":655},[97,1434,1436],{"class":99,"line":1435},138,[97,1437,1438],{}," # Calculate threshold\n",[97,1440,1442],{"class":99,"line":1441},139,[97,1443,1444],{}," model.eval()\n",[97,1446,1448],{"class":99,"line":1447},140,[97,1449,1450],{}," with torch.no_grad():\n",[97,1452,1454],{"class":99,"line":1453},141,[97,1455,1456],{}," recon = model(data_t).numpy()\n",[97,1458,1460],{"class":99,"line":1459},142,[97,1461,1462],{}," errors = np.mean((recon - X_norm) ** 2, axis=1)\n",[97,1464,1466],{"class":99,"line":1465},143,[97,1467,1468],{}," thresh = float(errors.mean() + THRESHOLD_SIGMA * errors.std())\n",[97,1470,1472],{"class":99,"line":1471},144,[97,1473,656],{"emptyLinePlaceholder":655},[97,1475,1477],{"class":99,"line":1476},145,[97,1478,1479],{}," with open(\"threshold.json\", \"w\") as f:\n",[97,1481,1483],{"class":99,"line":1482},146,[97,1484,1485],{}," json.dump({\"threshold\": thresh}, f, indent=2)\n",[97,1487,1489],{"class":99,"line":1488},147,[97,1490,1491],{}," print(f\" Threshold (mean+{THRESHOLD_SIGMA}σ): {thresh:.6f}\")\n",[97,1493,1495],{"class":99,"line":1494},148,[97,1496,1497],{}," print(\" Saved threshold.json\")\n",[97,1499,1501],{"class":99,"line":1500},149,[97,1502,656],{"emptyLinePlaceholder":655},[97,1504,1506],{"class":99,"line":1505},150,[97,1507,1508],{}," # Export to ONNX\n",[97,1510,1512],{"class":99,"line":1511},151,[97,1513,1514],{}," # Built manually to avoid version conflicts between PyTorch and ONNX exporters\n",[97,1516,1518],{"class":99,"line":1517},152,[97,1519,1520],{}," layers = [(\"encoder.0\",\"enc0\"),(\"encoder.2\",\"enc2\"),\n",[97,1522,1524],{"class":99,"line":1523},153,[97,1525,1526],{}," (\"decoder.0\",\"dec0\"),(\"decoder.2\",\"dec2\")]\n",[97,1528,1530],{"class":99,"line":1529},154,[97,1531,1532],{}," has_relu = [True, True, True, False]\n",[97,1534,1536],{"class":99,"line":1535},155,[97,1537,1538],{}," inits, nodes = [], []\n",[97,1540,1542],{"class":99,"line":1541},156,[97,1543,1544],{}," cur = \"features\"\n",[97,1546,1548],{"class":99,"line":1547},157,[97,1549,656],{"emptyLinePlaceholder":655},[97,1551,1553],{"class":99,"line":1552},158,[97,1554,1555],{}," for (prefix, tag), relu in zip(layers, has_relu):\n",[97,1557,1559],{"class":99,"line":1558},159,[97,1560,1561],{}," w = model.state_dict()[f\"{prefix}.weight\"].numpy().T.astype(np.float32)\n",[97,1563,1565],{"class":99,"line":1564},160,[97,1566,1567],{}," b = model.state_dict()[f\"{prefix}.bias\"].numpy().astype(np.float32)\n",[97,1569,1571],{"class":99,"line":1570},161,[97,1572,1573],{}," inits += [numpy_helper.from_array(w, name=f\"w_{tag}\"),\n",[97,1575,1577],{"class":99,"line":1576},162,[97,1578,1579],{}," numpy_helper.from_array(b, name=f\"b_{tag}\")]\n",[97,1581,1583],{"class":99,"line":1582},163,[97,1584,1585],{}," mm = f\"mm_{tag}\"; add = f\"add_{tag}\"\n",[97,1587,1589],{"class":99,"line":1588},164,[97,1590,1591],{}," nodes += [helper.make_node(\"MatMul\", [cur, f\"w_{tag}\"], [mm]),\n",[97,1593,1595],{"class":99,"line":1594},165,[97,1596,1597],{}," helper.make_node(\"Add\", [mm, f\"b_{tag}\"], [add])]\n",[97,1599,1601],{"class":99,"line":1600},166,[97,1602,1603],{}," cur = add\n",[97,1605,1607],{"class":99,"line":1606},167,[97,1608,1609],{}," if relu:\n",[97,1611,1613],{"class":99,"line":1612},168,[97,1614,1615],{}," r = f\"relu_{tag}\"\n",[97,1617,1619],{"class":99,"line":1618},169,[97,1620,1621],{}," nodes.append(helper.make_node(\"Relu\", [cur], [r]))\n",[97,1623,1625],{"class":99,"line":1624},170,[97,1626,1627],{}," cur = r\n",[97,1629,1631],{"class":99,"line":1630},171,[97,1632,656],{"emptyLinePlaceholder":655},[97,1634,1636],{"class":99,"line":1635},172,[97,1637,1638],{}," graph = helper.make_graph(\n",[97,1640,1642],{"class":99,"line":1641},173,[97,1643,1644],{}," nodes, \"autoencoder\",\n",[97,1646,1648],{"class":99,"line":1647},174,[97,1649,1650],{}," [helper.make_tensor_value_info(\"features\", TensorProto.FLOAT, [None, N])],\n",[97,1652,1654],{"class":99,"line":1653},175,[97,1655,1656],{}," [helper.make_tensor_value_info(cur, TensorProto.FLOAT, [None, N])],\n",[97,1658,1660],{"class":99,"line":1659},176,[97,1661,1662],{}," initializer=inits,\n",[97,1664,1666],{"class":99,"line":1665},177,[97,1667,1668],{}," )\n",[97,1670,1672],{"class":99,"line":1671},178,[97,1673,1674],{}," proto = helper.make_model(graph, opset_imports=[helper.make_opsetid(\"\", 11)])\n",[97,1676,1678],{"class":99,"line":1677},179,[97,1679,1680],{}," proto.ir_version = 7\n",[97,1682,1684],{"class":99,"line":1683},180,[97,1685,1686],{}," onnx.checker.check_model(proto)\n",[97,1688,1690],{"class":99,"line":1689},181,[97,1691,1692],{}," onnx.save(proto, \"motor_autoencoder.onnx\")\n",[97,1694,1696],{"class":99,"line":1695},182,[97,1697,1698],{}," print(\" Exported motor_autoencoder.onnx\")\n",[97,1700,1702],{"class":99,"line":1701},183,[97,1703,656],{"emptyLinePlaceholder":655},[97,1705,1707],{"class":99,"line":1706},184,[97,1708,1709],{}," # Sanity check\n",[97,1711,1713],{"class":99,"line":1712},185,[97,1714,1715],{}," sess = ort.InferenceSession(\"motor_autoencoder.onnx\")\n",[97,1717,1719],{"class":99,"line":1718},186,[97,1720,1721],{}," out = sess.run(None, {\"features\": X_norm[:5].astype(np.float32)})[0]\n",[97,1723,1725],{"class":99,"line":1724},187,[97,1726,1727],{}," mse = float(np.mean((out - X_norm[:5]) ** 2))\n",[97,1729,1731],{"class":99,"line":1730},188,[97,1732,1733],{}," print(f\"\\n Sanity MSE (5 normal samples): {mse:.6f} threshold: {thresh:.6f}\")\n",[97,1735,1737],{"class":99,"line":1736},189,[97,1738,1739],{}," if mse \u003C thresh:\n",[97,1741,1743],{"class":99,"line":1742},190,[97,1744,1745],{}," print(\" ✓ Model correct , normal data scores below threshold.\")\n",[97,1747,1749],{"class":99,"line":1748},191,[97,1750,1059],{},[97,1752,1754],{"class":99,"line":1753},192,[97,1755,1756],{}," print(\" ⚠ Sanity MSE above threshold , collect more data and retrain.\")\n",[97,1758,1760],{"class":99,"line":1759},193,[97,1761,656],{"emptyLinePlaceholder":655},[97,1763,1765],{"class":99,"line":1764},194,[97,1766,1767],{}," print(\"\\nDone. Copy these 3 files to your Node-RED server:\")\n",[97,1769,1771],{"class":99,"line":1770},195,[97,1772,1773],{}," print(\" motor_autoencoder.onnx scaler_params.json threshold.json\")\n",[97,1775,1777],{"class":99,"line":1776},196,[97,1778,656],{"emptyLinePlaceholder":655},[97,1780,1782],{"class":99,"line":1781},197,[97,1783,1784],{},"def handle_sigint(sig, frame):\n",[97,1786,1788],{"class":99,"line":1787},198,[97,1789,1790],{}," stop_flag[0] = True\n",[97,1792,1794],{"class":99,"line":1793},199,[97,1795,1796],{}," if len(training_data) \u003C MIN_WINDOWS:\n",[97,1798,1800],{"class":99,"line":1799},200,[97,1801,1802],{}," print(f\"\\n\\nNeed at least {MIN_WINDOWS} windows. Restart and collect longer.\")\n",[97,1804,1806],{"class":99,"line":1805},201,[97,1807,1808],{}," sys.exit(1)\n",[97,1810,1812],{"class":99,"line":1811},202,[97,1813,1814],{}," train_and_export()\n",[97,1816,1818],{"class":99,"line":1817},203,[97,1819,1820],{}," sys.exit(0)\n",[97,1822,1824],{"class":99,"line":1823},204,[97,1825,656],{"emptyLinePlaceholder":655},[97,1827,1829],{"class":99,"line":1828},205,[97,1830,1831],{},"signal.signal(signal.SIGINT, handle_sigint)\n",[97,1833,1835],{"class":99,"line":1834},206,[97,1836,656],{"emptyLinePlaceholder":655},[97,1838,1840],{"class":99,"line":1839},207,[97,1841,1842],{},"client = mqtt.Client(callback_api_version=CallbackAPIVersion.VERSION2,\n",[97,1844,1846],{"class":99,"line":1845},208,[97,1847,1848],{}," client_id=CLIENT_ID)\n",[97,1850,1852],{"class":99,"line":1851},209,[97,1853,1854],{},"client.username_pw_set(USERNAME, PASSWORD)\n",[97,1856,1858],{"class":99,"line":1857},210,[97,1859,1860],{},"client.on_connect = on_connect\n",[97,1862,1864],{"class":99,"line":1863},211,[97,1865,1866],{},"client.on_message = on_message\n",[97,1868,1870],{"class":99,"line":1869},212,[97,1871,1872],{},"client.connect(BROKER, PORT, keepalive=60)\n",[97,1874,1876],{"class":99,"line":1875},213,[97,1877,1878],{},"client.loop_forever()\n",[12,1880,1881,1882,1885],{},"Let it collect for 5–10 minutes (aim for 300+ windows), then press ",[295,1883,1884],{},"Ctrl+C once"," and wait. The script will train the model and export three files:",[1887,1888,1889,1896,1902],"ul",{},[1890,1891,1892,1895],"li",{},[59,1893,1894],{},"motor_autoencoder.onnx"," , the trained model in a portable, runtime-agnostic format",[1890,1897,1898,1901],{},[59,1899,1900],{},"scaler_params.json"," , the scaling parameters used to normalise input features",[1890,1903,1904,1906],{},[59,1905,613],{}," , the reconstruction error value above which a reading is flagged as anomalous",[290,1908,1909],{},[12,1910,1911,1914,1915,1918],{},[295,1912,1913],{},"Sanity check:"," Watch the output at the end. ",[59,1916,1917],{},"✓ Model correct"," means the model correctly scores normal data below the threshold. A warning means you should collect more data with the motor under its typical load and retrain.",[36,1920,1922],{"id":1921},"when-to-retrain","When to Retrain",[12,1924,1925],{},"The model captures what normal looks like at the time of training. Plan to retrain after any significant change to the motor's operating conditions: a maintenance overhaul, a change in load profile, a new mounting position, or seasonal temperature shifts that affect the vibration baseline. The process is identical, run the script again with the motor under its new normal conditions, replace the three output files, and restart the Node-RED flow.",[28,1927,1929],{"id":1928},"part-3-deploying-in-node-red","Part 3: Deploying in Node-RED",[12,1931,1932],{},"The model now knows what healthy looks like. This section builds the Node-RED flow that runs continuously, scores every incoming vibration batch in real time, and raises an alert the moment something shifts.",[36,1934,1936],{"id":1935},"installing-the-ai-nodes","Installing the AI Nodes",[12,1938,1939],{},"FlowFuse provides a dedicated AI nodes package for Node-RED that includes ONNX runtime support.",[290,1941,1942],{},[12,1943,1944,1947,1948,1953,1954,397],{},[295,1945,1946],{},"Note:"," These nodes are only available to FlowFuse users. If you don't have an account, ",[44,1949,1952],{"href":1950,"rel":1951},"https:\u002F\u002Fapp.flowfuse.com\u002Faccount\u002Fcreate",[48],"get started here"," and follow the steps to ",[44,1955,1957],{"href":1956},"\u002Fblog\u002F2025\u002F09\u002Finstalling-node-red\u002F","run the device agent",[1959,1960,1961,1967,1973],"ol",{},[1890,1962,1963,1964],{},"Open the Node-RED editor and go to ",[295,1965,1966],{},"Menu → Manage Palette",[1890,1968,1969,1970],{},"Search for ",[59,1971,1972],{},"@flowfuse-nodes\u002Fnr-ai-nodes",[1890,1974,1975,1976],{},"Click ",[295,1977,1978],{},"Install",[12,1980,1981,1982,1985],{},"Once installed, you will see new nodes in the palette under the FlowFuse AI category. This guide uses the ",[295,1983,1984],{},"onnx"," node.",[12,1987,1988],{},[1989,1990],"img",{"alt":1991,"src":1992,"title":1991},"FlowFuse AI nodes visible in the Node-RED palette under the FlowFuse AI category","\u002Fblog\u002F2026\u002F02\u002Fimages\u002Fai-nodes.png",[36,1994,1996],{"id":1995},"loading-the-model-files","Loading the Model Files",[12,1998,1999],{},"Place your three model files in the FlowFuse Device Agent directory before building the flow:",[52,2001,2003],{"className":353,"code":2002,"language":355,"meta":61,"style":61},"sudo mkdir -p \u002Fopt\u002Fflowfuse-device\u002Fmodels\nsudo cp motor_autoencoder.onnx \u002Fopt\u002Fflowfuse-device\u002Fmodels\u002F\nsudo cp scaler_params.json \u002Fopt\u002Fflowfuse-device\u002Fmodels\u002F\nsudo cp threshold.json \u002Fopt\u002Fflowfuse-device\u002Fmodels\u002F\n",[59,2004,2005,2019,2032,2043],{"__ignoreMap":61},[97,2006,2007,2010,2013,2016],{"class":99,"line":100},[97,2008,2009],{"class":362},"sudo",[97,2011,2012],{"class":126}," mkdir",[97,2014,2015],{"class":126}," -p",[97,2017,2018],{"class":126}," \u002Fopt\u002Fflowfuse-device\u002Fmodels\n",[97,2020,2021,2023,2026,2029],{"class":99,"line":107},[97,2022,2009],{"class":362},[97,2024,2025],{"class":126}," cp",[97,2027,2028],{"class":126}," motor_autoencoder.onnx",[97,2030,2031],{"class":126}," \u002Fopt\u002Fflowfuse-device\u002Fmodels\u002F\n",[97,2033,2034,2036,2038,2041],{"class":99,"line":135},[97,2035,2009],{"class":362},[97,2037,2025],{"class":126},[97,2039,2040],{"class":126}," scaler_params.json",[97,2042,2031],{"class":126},[97,2044,2045,2047,2049,2052],{"class":99,"line":153},[97,2046,2009],{"class":362},[97,2048,2025],{"class":126},[97,2050,2051],{"class":126}," threshold.json",[97,2053,2031],{"class":126},[36,2055,2057],{"id":2056},"building-the-inference-flow","Building the Inference Flow",[12,2059,2060],{},"The flow has five stages: receive the payload, extract features, scale and prepare, run inference, and score the result.",[12,2062,2063],{},[295,2064,2065],{},"1. Subscribe to MQTT",[12,2067,2068,2069,2072],{},"Add an ",[295,2070,2071],{},"mqtt-in"," node and configure it to connect to the same broker and topic used during training. Set the output to auto-detect so the JSON payload is parsed automatically.",[12,2074,2075,2076,2080],{},"If you are using the built-in FlowFuse MQTT broker, use the ",[44,2077,2079],{"href":2078},"\u002Fnode-red\u002Fflowfuse\u002Fmqtt\u002F","FlowFuse MQTT nodes"," , these connect automatically when dragged into the flow.",[12,2082,2083],{},[295,2084,2085],{},"2. Extract Features",[12,2087,2088,2089,2092,2093,2096,2097,2100,2101,2104],{},"Add a ",[295,2090,2091],{},"function"," node. In the ",[295,2094,2095],{},"Setup"," tab, add the module ",[59,2098,2099],{},"fs",". Then paste the following into the ",[295,2102,2103],{},"On Message"," tab:",[52,2106,2110],{"className":2107,"code":2108,"language":2109,"meta":61,"style":61},"language-javascript shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","function extractFeatures(sig) {\n const arr = sig.map(Number);\n const n = arr.length;\n const mean = arr.reduce((a, b) => a + b, 0) \u002F n;\n const std = Math.sqrt(arr.reduce((a, b) => a + (b - mean) ** 2, 0) \u002F n) + 1e-9;\n const absArr = arr.map(Math.abs);\n const rms = Math.sqrt(arr.reduce((a, b) => a + b * b, 0) \u002F n);\n const peak = Math.max(...absArr);\n const meanAbs = absArr.reduce((a, b) => a + b, 0) \u002F n + 1e-9;\n const meanSqrtAbs = absArr.reduce((a, b) => a + Math.sqrt(b), 0) \u002F n;\n return [\n rms, peak, Math.max(...arr) - Math.min(...arr),\n peak \u002F (rms + 1e-9),\n arr.reduce((a, b) => a + (b - mean) ** 2, 0) \u002F n,\n Math.sqrt(arr.reduce((a, b) => a + (b - mean) ** 2, 0) \u002F n),\n arr.reduce((a, b) => a + ((b - mean) \u002F std) ** 3, 0) \u002F n,\n arr.reduce((a, b) => a + ((b - mean) \u002F std) ** 4, 0) \u002F n - 3,\n rms \u002F meanAbs, peak \u002F meanAbs, peak \u002F (meanSqrtAbs ** 2 + 1e-9),\n ];\n}\n\n\u002F\u002F Scaler and threshold are cached in flow context after the first message.\n\u002F\u002F If you update the model files, restart the Node-RED flow to reload them.\nif (!flow.get('scaler')) {\n const sc = JSON.parse(fs.readFileSync('\u002Fopt\u002Fflowfuse-device\u002Fmodels\u002Fscaler_params.json'));\n const th = JSON.parse(fs.readFileSync('\u002Fopt\u002Fflowfuse-device\u002Fmodels\u002Fthreshold.json'));\n flow.set('scaler', sc);\n flow.set('threshold', th.threshold);\n}\n\nconst scaler = flow.get('scaler');\nconst CLIP = scaler.clip || 5.0;\nconst MIN_STD = 0.1;\n\nconst raw = [\n ...extractFeatures(msg.payload.x),\n ...extractFeatures(msg.payload.y),\n ...extractFeatures(msg.payload.z)\n];\n\nconst normalised = raw.map((v, i) => {\n const s = Math.max(scaler.std[i], MIN_STD);\n const n = (v - scaler.mean[i]) \u002F s;\n return Math.max(-CLIP, Math.min(CLIP, n));\n});\n\nmsg.input = {\n data: new Float32Array(normalised),\n type: \"float32\",\n dims: [1, 33]\n};\nmsg.payload = msg.input;\nmsg.threshold = flow.get('threshold');\nreturn msg;\n","javascript",[59,2111,2112,2132,2162,2180,2234,2317,2346,2408,2436,2487,2544,2553,2598,2619,2672,2735,2795,2858,2898,2905,2909,2913,2919,2924,2957,2997,3033,3059,3088,3092,3096,3126,3151,3165,3169,3180,3203,3222,3239,3246,3250,3284,3325,3359,3395,3404,3408,3422,3440,3456,3475,3480,3501,3530],{"__ignoreMap":61},[97,2113,2114,2116,2119,2122,2126,2129],{"class":99,"line":100},[97,2115,2091],{"class":113},[97,2117,2118],{"class":377}," extractFeatures",[97,2120,2121],{"class":103},"(",[97,2123,2125],{"class":2124},"sHdIc","sig",[97,2127,2128],{"class":103},")",[97,2130,2131],{"class":103}," {\n",[97,2133,2134,2137,2140,2143,2146,2148,2151,2154,2157,2159],{"class":99,"line":107},[97,2135,2136],{"class":113}," const",[97,2138,2139],{"class":189}," arr",[97,2141,2142],{"class":103}," =",[97,2144,2145],{"class":189}," sig",[97,2147,397],{"class":103},[97,2149,2150],{"class":377},"map",[97,2152,2121],{"class":2153},"swJcz",[97,2155,2156],{"class":189},"Number",[97,2158,2128],{"class":2153},[97,2160,2161],{"class":103},";\n",[97,2163,2164,2166,2169,2171,2173,2175,2178],{"class":99,"line":135},[97,2165,2136],{"class":113},[97,2167,2168],{"class":189}," n",[97,2170,2142],{"class":103},[97,2172,2139],{"class":189},[97,2174,397],{"class":103},[97,2176,2177],{"class":189},"length",[97,2179,2161],{"class":103},[97,2181,2182,2184,2187,2189,2191,2193,2196,2198,2200,2202,2204,2207,2209,2212,2215,2218,2220,2222,2225,2228,2230,2232],{"class":99,"line":153},[97,2183,2136],{"class":113},[97,2185,2186],{"class":189}," mean",[97,2188,2142],{"class":103},[97,2190,2139],{"class":189},[97,2192,397],{"class":103},[97,2194,2195],{"class":377},"reduce",[97,2197,2121],{"class":2153},[97,2199,2121],{"class":103},[97,2201,44],{"class":2124},[97,2203,171],{"class":103},[97,2205,2206],{"class":2124}," b",[97,2208,2128],{"class":103},[97,2210,2211],{"class":113}," =>",[97,2213,2214],{"class":189}," a",[97,2216,2217],{"class":103}," +",[97,2219,2206],{"class":189},[97,2221,171],{"class":103},[97,2223,2224],{"class":147}," 0",[97,2226,2227],{"class":2153},") ",[97,2229,316],{"class":103},[97,2231,2168],{"class":189},[97,2233,2161],{"class":103},[97,2235,2236,2238,2241,2243,2246,2248,2251,2253,2256,2258,2260,2262,2264,2266,2268,2270,2272,2274,2276,2278,2281,2284,2287,2289,2291,2294,2297,2299,2301,2303,2305,2307,2309,2312,2315],{"class":99,"line":196},[97,2237,2136],{"class":113},[97,2239,2240],{"class":189}," std",[97,2242,2142],{"class":103},[97,2244,2245],{"class":189}," Math",[97,2247,397],{"class":103},[97,2249,2250],{"class":377},"sqrt",[97,2252,2121],{"class":2153},[97,2254,2255],{"class":189},"arr",[97,2257,397],{"class":103},[97,2259,2195],{"class":377},[97,2261,2121],{"class":2153},[97,2263,2121],{"class":103},[97,2265,44],{"class":2124},[97,2267,171],{"class":103},[97,2269,2206],{"class":2124},[97,2271,2128],{"class":103},[97,2273,2211],{"class":113},[97,2275,2214],{"class":189},[97,2277,2217],{"class":103},[97,2279,2280],{"class":2153}," (",[97,2282,2283],{"class":189},"b",[97,2285,2286],{"class":103}," -",[97,2288,2186],{"class":189},[97,2290,2227],{"class":2153},[97,2292,2293],{"class":103},"**",[97,2295,2296],{"class":147}," 2",[97,2298,171],{"class":103},[97,2300,2224],{"class":147},[97,2302,2227],{"class":2153},[97,2304,316],{"class":103},[97,2306,2168],{"class":189},[97,2308,2227],{"class":2153},[97,2310,2311],{"class":103},"+",[97,2313,2314],{"class":147}," 1e-9",[97,2316,2161],{"class":103},[97,2318,2319,2321,2324,2326,2328,2330,2332,2334,2337,2339,2342,2344],{"class":99,"line":234},[97,2320,2136],{"class":113},[97,2322,2323],{"class":189}," absArr",[97,2325,2142],{"class":103},[97,2327,2139],{"class":189},[97,2329,397],{"class":103},[97,2331,2150],{"class":377},[97,2333,2121],{"class":2153},[97,2335,2336],{"class":189},"Math",[97,2338,397],{"class":103},[97,2340,2341],{"class":189},"abs",[97,2343,2128],{"class":2153},[97,2345,2161],{"class":103},[97,2347,2348,2350,2353,2355,2357,2359,2361,2363,2365,2367,2369,2371,2373,2375,2377,2379,2381,2383,2385,2387,2389,2392,2394,2396,2398,2400,2402,2404,2406],{"class":99,"line":273},[97,2349,2136],{"class":113},[97,2351,2352],{"class":189}," rms",[97,2354,2142],{"class":103},[97,2356,2245],{"class":189},[97,2358,397],{"class":103},[97,2360,2250],{"class":377},[97,2362,2121],{"class":2153},[97,2364,2255],{"class":189},[97,2366,397],{"class":103},[97,2368,2195],{"class":377},[97,2370,2121],{"class":2153},[97,2372,2121],{"class":103},[97,2374,44],{"class":2124},[97,2376,171],{"class":103},[97,2378,2206],{"class":2124},[97,2380,2128],{"class":103},[97,2382,2211],{"class":113},[97,2384,2214],{"class":189},[97,2386,2217],{"class":103},[97,2388,2206],{"class":189},[97,2390,2391],{"class":103}," *",[97,2393,2206],{"class":189},[97,2395,171],{"class":103},[97,2397,2224],{"class":147},[97,2399,2227],{"class":2153},[97,2401,316],{"class":103},[97,2403,2168],{"class":189},[97,2405,2128],{"class":2153},[97,2407,2161],{"class":103},[97,2409,2410,2412,2415,2417,2419,2421,2424,2426,2429,2432,2434],{"class":99,"line":674},[97,2411,2136],{"class":113},[97,2413,2414],{"class":189}," peak",[97,2416,2142],{"class":103},[97,2418,2245],{"class":189},[97,2420,397],{"class":103},[97,2422,2423],{"class":377},"max",[97,2425,2121],{"class":2153},[97,2427,2428],{"class":103},"...",[97,2430,2431],{"class":189},"absArr",[97,2433,2128],{"class":2153},[97,2435,2161],{"class":103},[97,2437,2438,2440,2443,2445,2447,2449,2451,2453,2455,2457,2459,2461,2463,2465,2467,2469,2471,2473,2475,2477,2479,2481,2483,2485],{"class":99,"line":680},[97,2439,2136],{"class":113},[97,2441,2442],{"class":189}," meanAbs",[97,2444,2142],{"class":103},[97,2446,2323],{"class":189},[97,2448,397],{"class":103},[97,2450,2195],{"class":377},[97,2452,2121],{"class":2153},[97,2454,2121],{"class":103},[97,2456,44],{"class":2124},[97,2458,171],{"class":103},[97,2460,2206],{"class":2124},[97,2462,2128],{"class":103},[97,2464,2211],{"class":113},[97,2466,2214],{"class":189},[97,2468,2217],{"class":103},[97,2470,2206],{"class":189},[97,2472,171],{"class":103},[97,2474,2224],{"class":147},[97,2476,2227],{"class":2153},[97,2478,316],{"class":103},[97,2480,2168],{"class":189},[97,2482,2217],{"class":103},[97,2484,2314],{"class":147},[97,2486,2161],{"class":103},[97,2488,2489,2491,2494,2496,2498,2500,2502,2504,2506,2508,2510,2512,2514,2516,2518,2520,2522,2524,2526,2528,2530,2532,2534,2536,2538,2540,2542],{"class":99,"line":686},[97,2490,2136],{"class":113},[97,2492,2493],{"class":189}," meanSqrtAbs",[97,2495,2142],{"class":103},[97,2497,2323],{"class":189},[97,2499,397],{"class":103},[97,2501,2195],{"class":377},[97,2503,2121],{"class":2153},[97,2505,2121],{"class":103},[97,2507,44],{"class":2124},[97,2509,171],{"class":103},[97,2511,2206],{"class":2124},[97,2513,2128],{"class":103},[97,2515,2211],{"class":113},[97,2517,2214],{"class":189},[97,2519,2217],{"class":103},[97,2521,2245],{"class":189},[97,2523,397],{"class":103},[97,2525,2250],{"class":377},[97,2527,2121],{"class":2153},[97,2529,2283],{"class":189},[97,2531,2128],{"class":2153},[97,2533,171],{"class":103},[97,2535,2224],{"class":147},[97,2537,2227],{"class":2153},[97,2539,316],{"class":103},[97,2541,2168],{"class":189},[97,2543,2161],{"class":103},[97,2545,2546,2550],{"class":99,"line":692},[97,2547,2549],{"class":2548},"s7zQu"," return",[97,2551,2552],{"class":2153}," [\n",[97,2554,2555,2558,2560,2562,2564,2566,2568,2570,2572,2574,2576,2578,2581,2583,2585,2588,2590,2592,2594,2596],{"class":99,"line":698},[97,2556,2557],{"class":189}," rms",[97,2559,171],{"class":103},[97,2561,2414],{"class":189},[97,2563,171],{"class":103},[97,2565,2245],{"class":189},[97,2567,397],{"class":103},[97,2569,2423],{"class":377},[97,2571,2121],{"class":2153},[97,2573,2428],{"class":103},[97,2575,2255],{"class":189},[97,2577,2227],{"class":2153},[97,2579,2580],{"class":103},"-",[97,2582,2245],{"class":189},[97,2584,397],{"class":103},[97,2586,2587],{"class":377},"min",[97,2589,2121],{"class":2153},[97,2591,2428],{"class":103},[97,2593,2255],{"class":189},[97,2595,2128],{"class":2153},[97,2597,132],{"class":103},[97,2599,2600,2603,2606,2608,2611,2613,2615,2617],{"class":99,"line":704},[97,2601,2602],{"class":189}," peak",[97,2604,2605],{"class":103}," \u002F",[97,2607,2280],{"class":2153},[97,2609,2610],{"class":189},"rms",[97,2612,2217],{"class":103},[97,2614,2314],{"class":147},[97,2616,2128],{"class":2153},[97,2618,132],{"class":103},[97,2620,2621,2624,2626,2628,2630,2632,2634,2636,2638,2640,2642,2644,2646,2648,2650,2652,2654,2656,2658,2660,2662,2664,2666,2668,2670],{"class":99,"line":710},[97,2622,2623],{"class":189}," arr",[97,2625,397],{"class":103},[97,2627,2195],{"class":377},[97,2629,2121],{"class":2153},[97,2631,2121],{"class":103},[97,2633,44],{"class":2124},[97,2635,171],{"class":103},[97,2637,2206],{"class":2124},[97,2639,2128],{"class":103},[97,2641,2211],{"class":113},[97,2643,2214],{"class":189},[97,2645,2217],{"class":103},[97,2647,2280],{"class":2153},[97,2649,2283],{"class":189},[97,2651,2286],{"class":103},[97,2653,2186],{"class":189},[97,2655,2227],{"class":2153},[97,2657,2293],{"class":103},[97,2659,2296],{"class":147},[97,2661,171],{"class":103},[97,2663,2224],{"class":147},[97,2665,2227],{"class":2153},[97,2667,316],{"class":103},[97,2669,2168],{"class":189},[97,2671,132],{"class":103},[97,2673,2674,2677,2679,2681,2683,2685,2687,2689,2691,2693,2695,2697,2699,2701,2703,2705,2707,2709,2711,2713,2715,2717,2719,2721,2723,2725,2727,2729,2731,2733],{"class":99,"line":716},[97,2675,2676],{"class":189}," Math",[97,2678,397],{"class":103},[97,2680,2250],{"class":377},[97,2682,2121],{"class":2153},[97,2684,2255],{"class":189},[97,2686,397],{"class":103},[97,2688,2195],{"class":377},[97,2690,2121],{"class":2153},[97,2692,2121],{"class":103},[97,2694,44],{"class":2124},[97,2696,171],{"class":103},[97,2698,2206],{"class":2124},[97,2700,2128],{"class":103},[97,2702,2211],{"class":113},[97,2704,2214],{"class":189},[97,2706,2217],{"class":103},[97,2708,2280],{"class":2153},[97,2710,2283],{"class":189},[97,2712,2286],{"class":103},[97,2714,2186],{"class":189},[97,2716,2227],{"class":2153},[97,2718,2293],{"class":103},[97,2720,2296],{"class":147},[97,2722,171],{"class":103},[97,2724,2224],{"class":147},[97,2726,2227],{"class":2153},[97,2728,316],{"class":103},[97,2730,2168],{"class":189},[97,2732,2128],{"class":2153},[97,2734,132],{"class":103},[97,2736,2737,2739,2741,2743,2745,2747,2749,2751,2753,2755,2757,2759,2761,2764,2766,2768,2770,2772,2774,2776,2778,2780,2783,2785,2787,2789,2791,2793],{"class":99,"line":722},[97,2738,2623],{"class":189},[97,2740,397],{"class":103},[97,2742,2195],{"class":377},[97,2744,2121],{"class":2153},[97,2746,2121],{"class":103},[97,2748,44],{"class":2124},[97,2750,171],{"class":103},[97,2752,2206],{"class":2124},[97,2754,2128],{"class":103},[97,2756,2211],{"class":113},[97,2758,2214],{"class":189},[97,2760,2217],{"class":103},[97,2762,2763],{"class":2153}," ((",[97,2765,2283],{"class":189},[97,2767,2286],{"class":103},[97,2769,2186],{"class":189},[97,2771,2227],{"class":2153},[97,2773,316],{"class":103},[97,2775,2240],{"class":189},[97,2777,2227],{"class":2153},[97,2779,2293],{"class":103},[97,2781,2782],{"class":147}," 3",[97,2784,171],{"class":103},[97,2786,2224],{"class":147},[97,2788,2227],{"class":2153},[97,2790,316],{"class":103},[97,2792,2168],{"class":189},[97,2794,132],{"class":103},[97,2796,2797,2799,2801,2803,2805,2807,2809,2811,2813,2815,2817,2819,2821,2823,2825,2827,2829,2831,2833,2835,2837,2839,2842,2844,2846,2848,2850,2852,2854,2856],{"class":99,"line":727},[97,2798,2623],{"class":189},[97,2800,397],{"class":103},[97,2802,2195],{"class":377},[97,2804,2121],{"class":2153},[97,2806,2121],{"class":103},[97,2808,44],{"class":2124},[97,2810,171],{"class":103},[97,2812,2206],{"class":2124},[97,2814,2128],{"class":103},[97,2816,2211],{"class":113},[97,2818,2214],{"class":189},[97,2820,2217],{"class":103},[97,2822,2763],{"class":2153},[97,2824,2283],{"class":189},[97,2826,2286],{"class":103},[97,2828,2186],{"class":189},[97,2830,2227],{"class":2153},[97,2832,316],{"class":103},[97,2834,2240],{"class":189},[97,2836,2227],{"class":2153},[97,2838,2293],{"class":103},[97,2840,2841],{"class":147}," 4",[97,2843,171],{"class":103},[97,2845,2224],{"class":147},[97,2847,2227],{"class":2153},[97,2849,316],{"class":103},[97,2851,2168],{"class":189},[97,2853,2286],{"class":103},[97,2855,2782],{"class":147},[97,2857,132],{"class":103},[97,2859,2860,2862,2864,2866,2868,2870,2872,2874,2876,2878,2880,2882,2885,2888,2890,2892,2894,2896],{"class":99,"line":733},[97,2861,2557],{"class":189},[97,2863,2605],{"class":103},[97,2865,2442],{"class":189},[97,2867,171],{"class":103},[97,2869,2414],{"class":189},[97,2871,2605],{"class":103},[97,2873,2442],{"class":189},[97,2875,171],{"class":103},[97,2877,2414],{"class":189},[97,2879,2605],{"class":103},[97,2881,2280],{"class":2153},[97,2883,2884],{"class":189},"meanSqrtAbs",[97,2886,2887],{"class":103}," **",[97,2889,2296],{"class":147},[97,2891,2217],{"class":103},[97,2893,2314],{"class":147},[97,2895,2128],{"class":2153},[97,2897,132],{"class":103},[97,2899,2900,2903],{"class":99,"line":739},[97,2901,2902],{"class":2153}," ]",[97,2904,2161],{"class":103},[97,2906,2907],{"class":99,"line":745},[97,2908,276],{"class":103},[97,2910,2911],{"class":99,"line":751},[97,2912,656],{"emptyLinePlaceholder":655},[97,2914,2915],{"class":99,"line":757},[97,2916,2918],{"class":2917},"sHwdD","\u002F\u002F Scaler and threshold are cached in flow context after the first message.\n",[97,2920,2921],{"class":99,"line":763},[97,2922,2923],{"class":2917},"\u002F\u002F If you update the model files, restart the Node-RED flow to reload them.\n",[97,2925,2926,2929,2931,2934,2937,2939,2942,2944,2947,2950,2952,2955],{"class":99,"line":769},[97,2927,2928],{"class":2548},"if",[97,2930,2280],{"class":189},[97,2932,2933],{"class":103},"!",[97,2935,2936],{"class":189},"flow",[97,2938,397],{"class":103},[97,2940,2941],{"class":377},"get",[97,2943,2121],{"class":189},[97,2945,2946],{"class":103},"'",[97,2948,2949],{"class":126},"scaler",[97,2951,2946],{"class":103},[97,2953,2954],{"class":189},")) ",[97,2956,104],{"class":103},[97,2958,2959,2961,2964,2966,2969,2971,2974,2976,2978,2980,2983,2985,2987,2990,2992,2995],{"class":99,"line":774},[97,2960,2136],{"class":113},[97,2962,2963],{"class":189}," sc",[97,2965,2142],{"class":103},[97,2967,2968],{"class":189}," JSON",[97,2970,397],{"class":103},[97,2972,2973],{"class":377},"parse",[97,2975,2121],{"class":2153},[97,2977,2099],{"class":189},[97,2979,397],{"class":103},[97,2981,2982],{"class":377},"readFileSync",[97,2984,2121],{"class":2153},[97,2986,2946],{"class":103},[97,2988,2989],{"class":126},"\u002Fopt\u002Fflowfuse-device\u002Fmodels\u002Fscaler_params.json",[97,2991,2946],{"class":103},[97,2993,2994],{"class":2153},"))",[97,2996,2161],{"class":103},[97,2998,2999,3001,3004,3006,3008,3010,3012,3014,3016,3018,3020,3022,3024,3027,3029,3031],{"class":99,"line":780},[97,3000,2136],{"class":113},[97,3002,3003],{"class":189}," th",[97,3005,2142],{"class":103},[97,3007,2968],{"class":189},[97,3009,397],{"class":103},[97,3011,2973],{"class":377},[97,3013,2121],{"class":2153},[97,3015,2099],{"class":189},[97,3017,397],{"class":103},[97,3019,2982],{"class":377},[97,3021,2121],{"class":2153},[97,3023,2946],{"class":103},[97,3025,3026],{"class":126},"\u002Fopt\u002Fflowfuse-device\u002Fmodels\u002Fthreshold.json",[97,3028,2946],{"class":103},[97,3030,2994],{"class":2153},[97,3032,2161],{"class":103},[97,3034,3035,3038,3040,3043,3045,3047,3049,3051,3053,3055,3057],{"class":99,"line":786},[97,3036,3037],{"class":189}," flow",[97,3039,397],{"class":103},[97,3041,3042],{"class":377},"set",[97,3044,2121],{"class":2153},[97,3046,2946],{"class":103},[97,3048,2949],{"class":126},[97,3050,2946],{"class":103},[97,3052,171],{"class":103},[97,3054,2963],{"class":189},[97,3056,2128],{"class":2153},[97,3058,2161],{"class":103},[97,3060,3061,3063,3065,3067,3069,3071,3074,3076,3078,3080,3082,3084,3086],{"class":99,"line":792},[97,3062,3037],{"class":189},[97,3064,397],{"class":103},[97,3066,3042],{"class":377},[97,3068,2121],{"class":2153},[97,3070,2946],{"class":103},[97,3072,3073],{"class":126},"threshold",[97,3075,2946],{"class":103},[97,3077,171],{"class":103},[97,3079,3003],{"class":189},[97,3081,397],{"class":103},[97,3083,3073],{"class":189},[97,3085,2128],{"class":2153},[97,3087,2161],{"class":103},[97,3089,3090],{"class":99,"line":798},[97,3091,276],{"class":103},[97,3093,3094],{"class":99,"line":804},[97,3095,656],{"emptyLinePlaceholder":655},[97,3097,3098,3101,3104,3107,3110,3112,3114,3116,3118,3120,3122,3124],{"class":99,"line":810},[97,3099,3100],{"class":113},"const",[97,3102,3103],{"class":189}," scaler ",[97,3105,3106],{"class":103},"=",[97,3108,3109],{"class":189}," flow",[97,3111,397],{"class":103},[97,3113,2941],{"class":377},[97,3115,2121],{"class":189},[97,3117,2946],{"class":103},[97,3119,2949],{"class":126},[97,3121,2946],{"class":103},[97,3123,2128],{"class":189},[97,3125,2161],{"class":103},[97,3127,3128,3130,3133,3135,3138,3140,3143,3146,3149],{"class":99,"line":816},[97,3129,3100],{"class":113},[97,3131,3132],{"class":189}," CLIP ",[97,3134,3106],{"class":103},[97,3136,3137],{"class":189}," scaler",[97,3139,397],{"class":103},[97,3141,3142],{"class":189},"clip ",[97,3144,3145],{"class":103},"||",[97,3147,3148],{"class":147}," 5.0",[97,3150,2161],{"class":103},[97,3152,3153,3155,3158,3160,3163],{"class":99,"line":822},[97,3154,3100],{"class":113},[97,3156,3157],{"class":189}," MIN_STD ",[97,3159,3106],{"class":103},[97,3161,3162],{"class":147}," 0.1",[97,3164,2161],{"class":103},[97,3166,3167],{"class":99,"line":827},[97,3168,656],{"emptyLinePlaceholder":655},[97,3170,3171,3173,3176,3178],{"class":99,"line":833},[97,3172,3100],{"class":113},[97,3174,3175],{"class":189}," raw ",[97,3177,3106],{"class":103},[97,3179,2552],{"class":189},[97,3181,3182,3185,3188,3191,3193,3196,3198,3201],{"class":99,"line":839},[97,3183,3184],{"class":103}," ...",[97,3186,3187],{"class":377},"extractFeatures",[97,3189,3190],{"class":189},"(msg",[97,3192,397],{"class":103},[97,3194,3195],{"class":189},"payload",[97,3197,397],{"class":103},[97,3199,3200],{"class":189},"x)",[97,3202,132],{"class":103},[97,3204,3205,3207,3209,3211,3213,3215,3217,3220],{"class":99,"line":844},[97,3206,3184],{"class":103},[97,3208,3187],{"class":377},[97,3210,3190],{"class":189},[97,3212,397],{"class":103},[97,3214,3195],{"class":189},[97,3216,397],{"class":103},[97,3218,3219],{"class":189},"y)",[97,3221,132],{"class":103},[97,3223,3224,3226,3228,3230,3232,3234,3236],{"class":99,"line":850},[97,3225,3184],{"class":103},[97,3227,3187],{"class":377},[97,3229,3190],{"class":189},[97,3231,397],{"class":103},[97,3233,3195],{"class":189},[97,3235,397],{"class":103},[97,3237,3238],{"class":189},"z)\n",[97,3240,3241,3244],{"class":99,"line":856},[97,3242,3243],{"class":189},"]",[97,3245,2161],{"class":103},[97,3247,3248],{"class":99,"line":862},[97,3249,656],{"emptyLinePlaceholder":655},[97,3251,3252,3254,3257,3259,3262,3264,3266,3268,3270,3273,3275,3278,3280,3282],{"class":99,"line":868},[97,3253,3100],{"class":113},[97,3255,3256],{"class":189}," normalised ",[97,3258,3106],{"class":103},[97,3260,3261],{"class":189}," raw",[97,3263,397],{"class":103},[97,3265,2150],{"class":377},[97,3267,2121],{"class":189},[97,3269,2121],{"class":103},[97,3271,3272],{"class":2124},"v",[97,3274,171],{"class":103},[97,3276,3277],{"class":2124}," i",[97,3279,2128],{"class":103},[97,3281,2211],{"class":113},[97,3283,2131],{"class":103},[97,3285,3286,3288,3291,3293,3295,3297,3299,3301,3303,3305,3308,3311,3314,3316,3318,3321,3323],{"class":99,"line":874},[97,3287,2136],{"class":113},[97,3289,3290],{"class":189}," s",[97,3292,2142],{"class":103},[97,3294,2245],{"class":189},[97,3296,397],{"class":103},[97,3298,2423],{"class":377},[97,3300,2121],{"class":2153},[97,3302,2949],{"class":189},[97,3304,397],{"class":103},[97,3306,3307],{"class":189},"std",[97,3309,3310],{"class":2153},"[",[97,3312,3313],{"class":189},"i",[97,3315,3243],{"class":2153},[97,3317,171],{"class":103},[97,3319,3320],{"class":189}," MIN_STD",[97,3322,2128],{"class":2153},[97,3324,2161],{"class":103},[97,3326,3327,3329,3331,3333,3335,3337,3339,3341,3343,3346,3348,3350,3353,3355,3357],{"class":99,"line":879},[97,3328,2136],{"class":113},[97,3330,2168],{"class":189},[97,3332,2142],{"class":103},[97,3334,2280],{"class":2153},[97,3336,3272],{"class":189},[97,3338,2286],{"class":103},[97,3340,3137],{"class":189},[97,3342,397],{"class":103},[97,3344,3345],{"class":189},"mean",[97,3347,3310],{"class":2153},[97,3349,3313],{"class":189},[97,3351,3352],{"class":2153},"]) ",[97,3354,316],{"class":103},[97,3356,3290],{"class":189},[97,3358,2161],{"class":103},[97,3360,3361,3363,3365,3367,3369,3371,3373,3375,3377,3379,3381,3383,3385,3387,3389,3391,3393],{"class":99,"line":885},[97,3362,2549],{"class":2548},[97,3364,2245],{"class":189},[97,3366,397],{"class":103},[97,3368,2423],{"class":377},[97,3370,2121],{"class":2153},[97,3372,2580],{"class":103},[97,3374,558],{"class":189},[97,3376,171],{"class":103},[97,3378,2245],{"class":189},[97,3380,397],{"class":103},[97,3382,2587],{"class":377},[97,3384,2121],{"class":2153},[97,3386,558],{"class":189},[97,3388,171],{"class":103},[97,3390,2168],{"class":189},[97,3392,2994],{"class":2153},[97,3394,2161],{"class":103},[97,3396,3397,3400,3402],{"class":99,"line":891},[97,3398,3399],{"class":103},"}",[97,3401,2128],{"class":189},[97,3403,2161],{"class":103},[97,3405,3406],{"class":99,"line":897},[97,3407,656],{"emptyLinePlaceholder":655},[97,3409,3410,3413,3415,3418,3420],{"class":99,"line":903},[97,3411,3412],{"class":189},"msg",[97,3414,397],{"class":103},[97,3416,3417],{"class":189},"input ",[97,3419,3106],{"class":103},[97,3421,2131],{"class":103},[97,3423,3424,3427,3429,3432,3435,3438],{"class":99,"line":909},[97,3425,3426],{"class":2153}," data",[97,3428,120],{"class":103},[97,3430,3431],{"class":103}," new",[97,3433,3434],{"class":377}," Float32Array",[97,3436,3437],{"class":189},"(normalised)",[97,3439,132],{"class":103},[97,3441,3442,3445,3447,3449,3452,3454],{"class":99,"line":915},[97,3443,3444],{"class":2153}," type",[97,3446,120],{"class":103},[97,3448,123],{"class":103},[97,3450,3451],{"class":126},"float32",[97,3453,117],{"class":103},[97,3455,132],{"class":103},[97,3457,3458,3461,3463,3465,3468,3470,3473],{"class":99,"line":921},[97,3459,3460],{"class":2153}," dims",[97,3462,120],{"class":103},[97,3464,165],{"class":189},[97,3466,3467],{"class":147},"1",[97,3469,171],{"class":103},[97,3471,3472],{"class":147}," 33",[97,3474,270],{"class":189},[97,3476,3477],{"class":99,"line":927},[97,3478,3479],{"class":103},"};\n",[97,3481,3482,3484,3486,3489,3491,3494,3496,3499],{"class":99,"line":933},[97,3483,3412],{"class":189},[97,3485,397],{"class":103},[97,3487,3488],{"class":189},"payload ",[97,3490,3106],{"class":103},[97,3492,3493],{"class":189}," msg",[97,3495,397],{"class":103},[97,3497,3498],{"class":189},"input",[97,3500,2161],{"class":103},[97,3502,3503,3505,3507,3510,3512,3514,3516,3518,3520,3522,3524,3526,3528],{"class":99,"line":939},[97,3504,3412],{"class":189},[97,3506,397],{"class":103},[97,3508,3509],{"class":189},"threshold ",[97,3511,3106],{"class":103},[97,3513,3109],{"class":189},[97,3515,397],{"class":103},[97,3517,2941],{"class":377},[97,3519,2121],{"class":189},[97,3521,2946],{"class":103},[97,3523,3073],{"class":126},[97,3525,2946],{"class":103},[97,3527,2128],{"class":189},[97,3529,2161],{"class":103},[97,3531,3532,3535,3537],{"class":99,"line":945},[97,3533,3534],{"class":2548},"return",[97,3536,3493],{"class":189},[97,3538,2161],{"class":103},[12,3540,3541,3542,3544,3545,3548],{},"This function extracts 11 time-domain features per axis (33 total), loads the scaler on first run, and normalises the feature vector. The ",[59,3543,548],{}," floor and ",[59,3546,3547],{},"±CLIP"," clamp mirror the values used during training and prevent near-constant features from producing extreme values, which are a common source of false positives with vibration sensors.",[290,3550,3551],{},[12,3552,3553,3556,3557,3559],{},[295,3554,3555],{},"If you changed the model file path",", update the two ",[59,3558,2982],{}," paths to match your chosen directory.",[12,3561,3562],{},[295,3563,3564],{},"3. Run the Model",[12,3566,2068,3567,3569],{},[295,3568,1984],{}," node and configure it:",[1887,3571,3572,3581],{},[1890,3573,3574,3577,3578],{},[295,3575,3576],{},"Model path:"," ",[59,3579,3580],{},"\u002Fopt\u002Fflowfuse-device\u002Fmodels\u002Fmotor_autoencoder.onnx",[1890,3582,3583,3577,3586],{},[295,3584,3585],{},"Input:",[59,3587,3588],{},"msg.payload",[12,3590,3591,3592,3594],{},"The autoencoder compresses the 33-feature input through the bottleneck and reconstructs it on the output side. The reconstructed tensor is accessible in the next node via ",[59,3593,3588],{},". Since the output key name depends on how the ONNX graph was exported, the scoring node retrieves it dynamically rather than relying on a hardcoded name.",[12,3596,3597],{},[295,3598,3599],{},"4. Score the Reconstruction Error",[12,3601,3602,3603,3605],{},"Add a second ",[295,3604,2091],{}," node:",[52,3607,3609],{"className":2107,"code":3608,"language":2109,"meta":61,"style":61},"if (!context.get('initialized')) {\n flow.set('score_history', []);\n context.set('initialized', true);\n}\n\n\u002F\u002F Generic output key lookup, works regardless of tensor name\nconst outputKey = Object.keys(msg.payload)[0];\nconst reconstructed = msg.payload[outputKey].cpuData;\n\nconst input = Array.from(msg.input.data);\nconst threshold = msg.threshold;\n\nconst mse = input.reduce((s, v, i) => s + (reconstructed[i] - v) ** 2, 0) \u002F input.length;\n\nlet history = flow.get('score_history') || [];\nhistory.push(mse);\nif (history.length > 10) history.shift();\nflow.set('score_history', history);\n\nconst smoothed = history.reduce((a, b) => a + b, 0) \u002F history.length;\n\nmsg.anomaly_score = smoothed;\nmsg.is_anomaly = smoothed > threshold;\nmsg.severity = smoothed > threshold * 2 ? 'CRITICAL' : smoothed > threshold ? 'WARNING' : 'NORMAL';\nmsg.payload = { anomaly_score: smoothed, threshold, is_anomaly: msg.is_anomaly, severity: msg.severity };\nreturn msg;\n",[59,3610,3611,3639,3663,3690,3694,3698,3703,3734,3757,3761,3791,3808,3812,3881,3885,3918,3933,3964,3987,3991,4044,4048,4064,4084,4147,4202],{"__ignoreMap":61},[97,3612,3613,3615,3617,3619,3622,3624,3626,3628,3630,3633,3635,3637],{"class":99,"line":100},[97,3614,2928],{"class":2548},[97,3616,2280],{"class":189},[97,3618,2933],{"class":103},[97,3620,3621],{"class":189},"context",[97,3623,397],{"class":103},[97,3625,2941],{"class":377},[97,3627,2121],{"class":189},[97,3629,2946],{"class":103},[97,3631,3632],{"class":126},"initialized",[97,3634,2946],{"class":103},[97,3636,2954],{"class":189},[97,3638,104],{"class":103},[97,3640,3641,3643,3645,3647,3649,3651,3654,3656,3658,3661],{"class":99,"line":107},[97,3642,3037],{"class":189},[97,3644,397],{"class":103},[97,3646,3042],{"class":377},[97,3648,2121],{"class":2153},[97,3650,2946],{"class":103},[97,3652,3653],{"class":126},"score_history",[97,3655,2946],{"class":103},[97,3657,171],{"class":103},[97,3659,3660],{"class":2153}," [])",[97,3662,2161],{"class":103},[97,3664,3665,3668,3670,3672,3674,3676,3678,3680,3682,3686,3688],{"class":99,"line":135},[97,3666,3667],{"class":189}," context",[97,3669,397],{"class":103},[97,3671,3042],{"class":377},[97,3673,2121],{"class":2153},[97,3675,2946],{"class":103},[97,3677,3632],{"class":126},[97,3679,2946],{"class":103},[97,3681,171],{"class":103},[97,3683,3685],{"class":3684},"sfNiH"," true",[97,3687,2128],{"class":2153},[97,3689,2161],{"class":103},[97,3691,3692],{"class":99,"line":153},[97,3693,276],{"class":103},[97,3695,3696],{"class":99,"line":196},[97,3697,656],{"emptyLinePlaceholder":655},[97,3699,3700],{"class":99,"line":234},[97,3701,3702],{"class":2917},"\u002F\u002F Generic output key lookup, works regardless of tensor name\n",[97,3704,3705,3707,3710,3712,3715,3717,3720,3722,3724,3727,3730,3732],{"class":99,"line":273},[97,3706,3100],{"class":113},[97,3708,3709],{"class":189}," outputKey ",[97,3711,3106],{"class":103},[97,3713,3714],{"class":189}," Object",[97,3716,397],{"class":103},[97,3718,3719],{"class":377},"keys",[97,3721,3190],{"class":189},[97,3723,397],{"class":103},[97,3725,3726],{"class":189},"payload)[",[97,3728,3729],{"class":147},"0",[97,3731,3243],{"class":189},[97,3733,2161],{"class":103},[97,3735,3736,3738,3741,3743,3745,3747,3750,3752,3755],{"class":99,"line":674},[97,3737,3100],{"class":113},[97,3739,3740],{"class":189}," reconstructed ",[97,3742,3106],{"class":103},[97,3744,3493],{"class":189},[97,3746,397],{"class":103},[97,3748,3749],{"class":189},"payload[outputKey]",[97,3751,397],{"class":103},[97,3753,3754],{"class":189},"cpuData",[97,3756,2161],{"class":103},[97,3758,3759],{"class":99,"line":680},[97,3760,656],{"emptyLinePlaceholder":655},[97,3762,3763,3765,3768,3770,3773,3775,3778,3780,3782,3784,3786,3789],{"class":99,"line":686},[97,3764,3100],{"class":113},[97,3766,3767],{"class":189}," input ",[97,3769,3106],{"class":103},[97,3771,3772],{"class":189}," Array",[97,3774,397],{"class":103},[97,3776,3777],{"class":377},"from",[97,3779,3190],{"class":189},[97,3781,397],{"class":103},[97,3783,3498],{"class":189},[97,3785,397],{"class":103},[97,3787,3788],{"class":189},"data)",[97,3790,2161],{"class":103},[97,3792,3793,3795,3798,3800,3802,3804,3806],{"class":99,"line":692},[97,3794,3100],{"class":113},[97,3796,3797],{"class":189}," threshold ",[97,3799,3106],{"class":103},[97,3801,3493],{"class":189},[97,3803,397],{"class":103},[97,3805,3073],{"class":189},[97,3807,2161],{"class":103},[97,3809,3810],{"class":99,"line":698},[97,3811,656],{"emptyLinePlaceholder":655},[97,3813,3814,3816,3819,3821,3824,3826,3828,3830,3832,3835,3837,3840,3842,3844,3846,3848,3851,3853,3856,3858,3861,3863,3865,3867,3869,3871,3873,3875,3877,3879],{"class":99,"line":704},[97,3815,3100],{"class":113},[97,3817,3818],{"class":189}," mse ",[97,3820,3106],{"class":103},[97,3822,3823],{"class":189}," input",[97,3825,397],{"class":103},[97,3827,2195],{"class":377},[97,3829,2121],{"class":189},[97,3831,2121],{"class":103},[97,3833,3834],{"class":2124},"s",[97,3836,171],{"class":103},[97,3838,3839],{"class":2124}," v",[97,3841,171],{"class":103},[97,3843,3277],{"class":2124},[97,3845,2128],{"class":103},[97,3847,2211],{"class":113},[97,3849,3850],{"class":189}," s ",[97,3852,2311],{"class":103},[97,3854,3855],{"class":189}," (reconstructed[i] ",[97,3857,2580],{"class":103},[97,3859,3860],{"class":189}," v) ",[97,3862,2293],{"class":103},[97,3864,2296],{"class":147},[97,3866,171],{"class":103},[97,3868,2224],{"class":147},[97,3870,2227],{"class":189},[97,3872,316],{"class":103},[97,3874,3823],{"class":189},[97,3876,397],{"class":103},[97,3878,2177],{"class":189},[97,3880,2161],{"class":103},[97,3882,3883],{"class":99,"line":710},[97,3884,656],{"emptyLinePlaceholder":655},[97,3886,3887,3890,3893,3895,3897,3899,3901,3903,3905,3907,3909,3911,3913,3916],{"class":99,"line":716},[97,3888,3889],{"class":113},"let",[97,3891,3892],{"class":189}," history ",[97,3894,3106],{"class":103},[97,3896,3109],{"class":189},[97,3898,397],{"class":103},[97,3900,2941],{"class":377},[97,3902,2121],{"class":189},[97,3904,2946],{"class":103},[97,3906,3653],{"class":126},[97,3908,2946],{"class":103},[97,3910,2227],{"class":189},[97,3912,3145],{"class":103},[97,3914,3915],{"class":189}," []",[97,3917,2161],{"class":103},[97,3919,3920,3923,3925,3928,3931],{"class":99,"line":722},[97,3921,3922],{"class":189},"history",[97,3924,397],{"class":103},[97,3926,3927],{"class":377},"push",[97,3929,3930],{"class":189},"(mse)",[97,3932,2161],{"class":103},[97,3934,3935,3937,3940,3942,3945,3948,3951,3954,3956,3959,3962],{"class":99,"line":727},[97,3936,2928],{"class":2548},[97,3938,3939],{"class":189}," (history",[97,3941,397],{"class":103},[97,3943,3944],{"class":189},"length ",[97,3946,3947],{"class":103},">",[97,3949,3950],{"class":147}," 10",[97,3952,3953],{"class":189},") history",[97,3955,397],{"class":103},[97,3957,3958],{"class":377},"shift",[97,3960,3961],{"class":189},"()",[97,3963,2161],{"class":103},[97,3965,3966,3968,3970,3972,3974,3976,3978,3980,3982,3985],{"class":99,"line":733},[97,3967,2936],{"class":189},[97,3969,397],{"class":103},[97,3971,3042],{"class":377},[97,3973,2121],{"class":189},[97,3975,2946],{"class":103},[97,3977,3653],{"class":126},[97,3979,2946],{"class":103},[97,3981,171],{"class":103},[97,3983,3984],{"class":189}," history)",[97,3986,2161],{"class":103},[97,3988,3989],{"class":99,"line":739},[97,3990,656],{"emptyLinePlaceholder":655},[97,3992,3993,3995,3998,4000,4003,4005,4007,4009,4011,4013,4015,4017,4019,4021,4024,4026,4028,4030,4032,4034,4036,4038,4040,4042],{"class":99,"line":745},[97,3994,3100],{"class":113},[97,3996,3997],{"class":189}," smoothed ",[97,3999,3106],{"class":103},[97,4001,4002],{"class":189}," history",[97,4004,397],{"class":103},[97,4006,2195],{"class":377},[97,4008,2121],{"class":189},[97,4010,2121],{"class":103},[97,4012,44],{"class":2124},[97,4014,171],{"class":103},[97,4016,2206],{"class":2124},[97,4018,2128],{"class":103},[97,4020,2211],{"class":113},[97,4022,4023],{"class":189}," a ",[97,4025,2311],{"class":103},[97,4027,2206],{"class":189},[97,4029,171],{"class":103},[97,4031,2224],{"class":147},[97,4033,2227],{"class":189},[97,4035,316],{"class":103},[97,4037,4002],{"class":189},[97,4039,397],{"class":103},[97,4041,2177],{"class":189},[97,4043,2161],{"class":103},[97,4045,4046],{"class":99,"line":751},[97,4047,656],{"emptyLinePlaceholder":655},[97,4049,4050,4052,4054,4057,4059,4062],{"class":99,"line":757},[97,4051,3412],{"class":189},[97,4053,397],{"class":103},[97,4055,4056],{"class":189},"anomaly_score ",[97,4058,3106],{"class":103},[97,4060,4061],{"class":189}," smoothed",[97,4063,2161],{"class":103},[97,4065,4066,4068,4070,4073,4075,4077,4079,4082],{"class":99,"line":763},[97,4067,3412],{"class":189},[97,4069,397],{"class":103},[97,4071,4072],{"class":189},"is_anomaly ",[97,4074,3106],{"class":103},[97,4076,3997],{"class":189},[97,4078,3947],{"class":103},[97,4080,4081],{"class":189}," threshold",[97,4083,2161],{"class":103},[97,4085,4086,4088,4090,4093,4095,4097,4099,4101,4104,4106,4109,4112,4115,4117,4120,4122,4124,4126,4129,4131,4134,4136,4138,4140,4143,4145],{"class":99,"line":769},[97,4087,3412],{"class":189},[97,4089,397],{"class":103},[97,4091,4092],{"class":189},"severity ",[97,4094,3106],{"class":103},[97,4096,3997],{"class":189},[97,4098,3947],{"class":103},[97,4100,3797],{"class":189},[97,4102,4103],{"class":103},"*",[97,4105,2296],{"class":147},[97,4107,4108],{"class":103}," ?",[97,4110,4111],{"class":103}," '",[97,4113,4114],{"class":126},"CRITICAL",[97,4116,2946],{"class":103},[97,4118,4119],{"class":103}," :",[97,4121,3997],{"class":189},[97,4123,3947],{"class":103},[97,4125,3797],{"class":189},[97,4127,4128],{"class":103},"?",[97,4130,4111],{"class":103},[97,4132,4133],{"class":126},"WARNING",[97,4135,2946],{"class":103},[97,4137,4119],{"class":103},[97,4139,4111],{"class":103},[97,4141,4142],{"class":126},"NORMAL",[97,4144,2946],{"class":103},[97,4146,2161],{"class":103},[97,4148,4149,4151,4153,4156,4158,4161,4164,4166,4168,4170,4172,4174,4177,4179,4181,4183,4186,4188,4191,4193,4195,4197,4200],{"class":99,"line":774},[97,4150,3412],{"class":189},[97,4152,397],{"class":103},[97,4154,4155],{"class":189},"payload ",[97,4157,3106],{"class":103},[97,4159,4160],{"class":103}," {",[97,4162,4163],{"class":2153}," anomaly_score",[97,4165,120],{"class":103},[97,4167,4061],{"class":189},[97,4169,171],{"class":103},[97,4171,4081],{"class":189},[97,4173,171],{"class":103},[97,4175,4176],{"class":2153}," is_anomaly",[97,4178,120],{"class":103},[97,4180,3493],{"class":189},[97,4182,397],{"class":103},[97,4184,4185],{"class":189},"is_anomaly",[97,4187,171],{"class":103},[97,4189,4190],{"class":2153}," severity",[97,4192,120],{"class":103},[97,4194,3493],{"class":189},[97,4196,397],{"class":103},[97,4198,4199],{"class":189},"severity ",[97,4201,3479],{"class":103},[97,4203,4204,4206,4208],{"class":99,"line":780},[97,4205,3534],{"class":2548},[97,4207,3493],{"class":189},[97,4209,2161],{"class":103},[12,4211,4212,4213,4215,4216,4218,4219,397],{},"This function computes mean squared error between the model's reconstruction and the normalised input, applies a 10-window rolling average to reduce sensitivity to transient spikes, then classifies the result as ",[59,4214,4142],{},", ",[59,4217,4133],{},", or ",[59,4220,4114],{},[290,4222,4223],{},[12,4224,4225,4228],{},[295,4226,4227],{},"Recovery time:"," After an anomaly clears, the score returns to normal once the rolling window fills with healthy readings , typically 10 × your publish interval. At 500 ms publishing, that's roughly 5 seconds. Reduce the history window size for faster recovery; increase it to suppress false alarms.",[12,4230,4231],{},"Once deployed, the flow should look like this:",[12,4233,4234],{},[1989,4235],{"alt":4236,"src":4237,"title":4236},"Completed Node-RED inference flow showing MQTT input, feature extraction function node, ONNX node, and anomaly scoring function node","\u002Fblog\u002F2026\u002F02\u002Fimages\u002Fflow.png",[12,4239,4240],{},[295,4241,4242],{},"5. Act on the Result",[12,4244,4245,4246,4251,4252,4215,4256,4218,4260,4264],{},"Connect the scoring output to whatever suits your operation. For testing, a debug node shows results in real time. For production, an mqtt-out node can publish anomaly alerts downstream, the ",[44,4247,4250],{"href":4248,"rel":4249},"https:\u002F\u002Fdashboard.flowfuse.com",[48],"FlowFuse Dashboard"," package can visualise the anomaly score over time with a clear motor state indicator, and alerting nodes can notify your team directly, via ",[44,4253,4255],{"href":4254},"\u002Fnode-red\u002Fnotification\u002Femail\u002F","Telegram",[44,4257,4259],{"href":4258},"\u002Fnode-red\u002Fnotification\u002Ftelegram\u002F","email",[44,4261,4263],{"href":4262},"\u002Fblueprints\u002Fother\u002Fmobile-alerting\u002F","SIGNL4"," for structured mobile alerts with on-call scheduling and acknowledgement tracking.",[36,4266,4268],{"id":4267},"what-the-output-looks-like","What the Output Looks Like",[12,4270,4271],{},"Each message produces a structured result:",[52,4273,4275],{"className":91,"code":4274,"language":93,"meta":61,"style":61},"{\n \"anomaly_score\": 0.842,\n \"threshold\": 0.703,\n \"is_anomaly\": true,\n \"severity\": \"WARNING\"\n}\n",[59,4276,4277,4281,4297,4312,4325,4343],{"__ignoreMap":61},[97,4278,4279],{"class":99,"line":100},[97,4280,104],{"class":103},[97,4282,4283,4285,4288,4290,4292,4295],{"class":99,"line":107},[97,4284,110],{"class":103},[97,4286,4287],{"class":113},"anomaly_score",[97,4289,117],{"class":103},[97,4291,120],{"class":103},[97,4293,4294],{"class":147}," 0.842",[97,4296,132],{"class":103},[97,4298,4299,4301,4303,4305,4307,4310],{"class":99,"line":135},[97,4300,110],{"class":103},[97,4302,3073],{"class":113},[97,4304,117],{"class":103},[97,4306,120],{"class":103},[97,4308,4309],{"class":147}," 0.703",[97,4311,132],{"class":103},[97,4313,4314,4316,4318,4320,4322],{"class":99,"line":153},[97,4315,110],{"class":103},[97,4317,4185],{"class":113},[97,4319,117],{"class":103},[97,4321,120],{"class":103},[97,4323,4324],{"class":103}," true,\n",[97,4326,4327,4329,4332,4334,4336,4338,4340],{"class":99,"line":196},[97,4328,110],{"class":103},[97,4330,4331],{"class":113},"severity",[97,4333,117],{"class":103},[97,4335,120],{"class":103},[97,4337,123],{"class":103},[97,4339,4133],{"class":126},[97,4341,4342],{"class":103},"\"\n",[97,4344,4345],{"class":99,"line":234},[97,4346,276],{"class":103},[12,4348,4349,4350,4352,4353,4356,4357,4360,4361,4363],{},"When ",[59,4351,4185],{}," is ",[59,4354,4355],{},"false",", the motor is behaving within the expected range. When it flips to ",[59,4358,4359],{},"true",", the vibration pattern has shifted beyond the acceptable boundary, giving you time to act before the problem becomes a failure. A severity of ",[59,4362,4114],{}," means the score has crossed twice the threshold, signalling a more significant deviation that warrants immediate attention.",[28,4365,4367],{"id":4366},"what-this-system-wont-tell-you","What This System Won't Tell You",[12,4369,4370],{},"This approach works well, but it's worth being clear about where it stops.",[12,4372,4373],{},"The autoencoder learns a statistical boundary around the vibration patterns it was trained on. It doesn't understand physics, it doesn't know the difference between a worn bearing and a loose mounting bolt, and it has no concept of severity beyond the reconstruction error score. When it flags an anomaly, it's telling you that something has changed, not what changed or why. Diagnosing the root cause still requires a technician with domain knowledge.",[12,4375,4376],{},"Training data quality matters more than model architecture. A model trained on data collected while the motor was lightly loaded, recently serviced, or running in cool ambient conditions will treat those as \"normal.\" If real operating conditions differ, the threshold may be poorly calibrated from day one, generating either chronic false positives or, worse, missing genuine faults. There's no substitute for collecting training data under representative, sustained, real-world load.",[12,4378,4379,4380,4382],{},"False positives are inevitable in early deployment. External vibration from nearby equipment, transient load spikes, or sensor cable movement can all push the score above threshold momentarily. The rolling average window helps, but it doesn't eliminate them. Treat the first few weeks as a calibration period: log alerts, investigate them, and adjust ",[59,4381,588],{}," or the window size based on what you learn. The system improves with attention.",[12,4384,4385],{},"Finally, anomaly detection is an early warning layer, not a maintenance strategy on its own. It tells you to look sooner, not what to do when you get there. Pair it with regular physical inspection, lubrication schedules, and, where possible, a domain expert who can interpret the alerts in context. Used that way, it earns its place. Used as a replacement for those things, it will eventually let you down.",[4387,4388,4389],"style",{},"html pre.shiki code .sMK4o, html code.shiki .sMK4o{--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF}html pre.shiki code .spNyl, html code.shiki .spNyl{--shiki-light:#9C3EDA;--shiki-default:#C792EA;--shiki-dark:#C792EA}html pre.shiki code .sfazB, html code.shiki .sfazB{--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D}html pre.shiki code .sbssI, html code.shiki .sbssI{--shiki-light:#F76D47;--shiki-default:#F78C6C;--shiki-dark:#F78C6C}html pre.shiki code .sTEyZ, html code.shiki .sTEyZ{--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8}html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html pre.shiki code .sBMFI, html code.shiki .sBMFI{--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B}html pre.shiki code .s2Zo4, html code.shiki .s2Zo4{--shiki-light:#6182B8;--shiki-default:#82AAFF;--shiki-dark:#82AAFF}html pre.shiki code .sHdIc, html code.shiki .sHdIc{--shiki-light:#90A4AE;--shiki-light-font-style:italic;--shiki-default:#EEFFFF;--shiki-default-font-style:italic;--shiki-dark:#BABED8;--shiki-dark-font-style:italic}html pre.shiki code .swJcz, html code.shiki .swJcz{--shiki-light:#E53935;--shiki-default:#F07178;--shiki-dark:#F07178}html pre.shiki code .s7zQu, html code.shiki .s7zQu{--shiki-light:#39ADB5;--shiki-light-font-style:italic;--shiki-default:#89DDFF;--shiki-default-font-style:italic;--shiki-dark:#89DDFF;--shiki-dark-font-style:italic}html pre.shiki code .sHwdD, html code.shiki .sHwdD{--shiki-light:#90A4AE;--shiki-light-font-style:italic;--shiki-default:#546E7A;--shiki-default-font-style:italic;--shiki-dark:#676E95;--shiki-dark-font-style:italic}html pre.shiki code .sfNiH, html code.shiki .sfNiH{--shiki-light:#FF5370;--shiki-default:#FF9CAC;--shiki-dark:#FF9CAC}",{"title":61,"searchDepth":153,"depth":153,"links":4391},[4392,4395,4396,4400,4406,4412],{"id":30,"depth":107,"text":31,"children":4393},[4394],{"id":38,"depth":135,"text":39},{"id":70,"depth":107,"text":71},{"id":77,"depth":107,"text":78,"children":4397},[4398,4399],{"id":84,"depth":135,"text":85},{"id":301,"depth":135,"text":302},{"id":333,"depth":107,"text":334,"children":4401},[4402,4403,4404,4405],{"id":340,"depth":135,"text":341},{"id":434,"depth":135,"text":435},{"id":617,"depth":135,"text":618},{"id":1921,"depth":135,"text":1922},{"id":1928,"depth":107,"text":1929,"children":4407},[4408,4409,4410,4411],{"id":1935,"depth":135,"text":1936},{"id":1995,"depth":135,"text":1996},{"id":2056,"depth":135,"text":2057},{"id":4267,"depth":135,"text":4268},{"id":4366,"depth":107,"text":4367},{"type":4414,"title":4415,"description":4416},"demo","See How FlowFuse Catches Faults Before They Fail","Book a demo and see how FlowFuse brings sensor ingestion, AI inference, and real-time alerting together on the factory floor, no separate ML infrastructure required","2026-02-20","Learn how to monitor industrial motors continuously, train a custom autoencoder on healthy vibration data, and deploy real-time anomaly detection in Node-RED.","md",null,"\u002Fblog\u002F2026\u002F02\u002Fimages\u002Fmotor-anomaly-detection-ai.png","2026-06-19",{"excerpt":4424},{"type":9,"value":4425},[4426],[12,4427,14],{},"\u002Fblog\u002F2026\u002F02\u002Fmotor-anomaly-detector-ai",{"title":5,"description":4418},{"loc":4428},"blog\u002F2026\u002F02\u002Fmotor-anomaly-detector-ai","Detect motor faults early using AI-driven vibration analysis and anomaly detection.",[4434,4435],"flowfuse","ai","This guide walks through building an AI-based motor vibration anomaly detector using an autoencoder trained on healthy sensor data, exported to ONNX, and deployed directly in Node-RED via FlowFuse's AI nodes no separate ML infrastructure required. The system extracts 33 time-domain features per vibration window, scores each reading against a trained threshold, and classifies results as NORMAL, WARNING, or CRITICAL in real time.","tgMGTSLEf3Sx1xT41PJZe5wYLeSFzAaFfIjA7jcV3kE",{"id":4439,"extension":4440,"meta":4441,"sections":4442,"stem":4829,"__hash__":4830},"featureCatalog\u002Ffeature-catalog.yml","yml",{},[4443,4494,4595,4655,4732,4811],{"id":4444,"title":4445,"features":4446},"ai-automation","AI & Automation",[4447,4457,4467,4477,4483,4489],{"id":4448,"title":4449,"description":4450,"docsLink":4451,"changelog":4452,"tiers":4456},"flowfuse-expert-ai","FlowFuse Expert AI","Build industrial apps with agents, and query the state of the factory with an agent. Adapt your current hardware and machines so agentic work can be done against them, without ripping and replacing what's already on the plant floor.","\u002Fdocs\u002Fuser\u002Fexpert\u002F",[4453],{"url":4454,"release":4455},"\u002Fchangelog\u002F2026\u002F02\u002Fff-expert-update-banner\u002F","2.28",{"edge":655,"hub":655,"fleet":655},{"id":4458,"title":4459,"description":4460,"docsLink":4461,"changelog":4462,"subfeature":655,"showOnPricing":4465,"tiers":4466},"flowfuse-expert-support-mode","Support Mode","Chat-based assistance for FlowFuse and Node-RED, including Node-RED instance management through natural language.","\u002Fdocs\u002Fuser\u002Fexpert\u002Fchat\u002F#support-mode",[4463],{"url":4464,"release":4455},"\u002Fchangelog\u002F2026\u002F02\u002Fff-expert-debug-log-context\u002F",false,{"edge":655,"hub":655,"fleet":655},{"id":4468,"title":4469,"description":4470,"docsLink":4471,"changelog":4472,"subfeature":655,"showOnPricing":4465,"tiers":4476},"flowfuse-expert-application-building","Application Building","Describe what you want to build and FlowFuse Expert assembles it on your workspace, adding tabs, wiring nodes, and configuring properties.","\u002Fdocs\u002Fuser\u002Fexpert\u002Fchat\u002F",[4473],{"url":4474,"release":4475},"\u002Fchangelog\u002F2026\u002F05\u002Fexpert-application-building\u002F","2.30",{"edge":655,"hub":655,"fleet":655},{"id":4478,"title":4479,"description":4480,"docsLink":4481,"subfeature":655,"beta":655,"showOnPricing":4465,"tiers":4482},"flowfuse-expert-insights-mode","Insights Mode","Connects FlowFuse Expert to MCP servers in your Node-RED instances, enabling real-time data queries and actions through a single chat interface.","\u002Fdocs\u002Fuser\u002Fexpert\u002Fchat\u002F#insights-mode",{"edge":655,"hub":655,"fleet":655},{"id":4484,"title":4485,"description":4486,"docsLink":4487,"tiers":4488},"mcp-servers","Agentic Operations","Expose your Node-RED flows as tools an AI agent can call directly, so agents can query the state of the factory or trigger actions without custom integration work.","\u002Fnode-red\u002Fflowfuse\u002Fmcp\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4490,"title":4491,"description":4492,"tiers":4493},"onnx-integration","ONNX Integration","Run trained machine learning models directly in your flows, including on edge hardware, without sending data out to an external inference service.",{"edge":655,"hub":655,"fleet":655},{"id":4495,"title":4496,"features":4497},"build","Build",[4498,4504,4510,4520,4526,4532,4536,4542,4548,4556,4562,4566,4570,4579,4587],{"id":4499,"title":4500,"description":4501,"docsLink":4502,"tiers":4503},"edge-development","Edge Development","Develop and test Node-RED flows directly on edge devices with a remote editor proxy.","\u002Fdocs\u002Fdevice-agent\u002Fquickstart\u002F",{"edge":655,"hub":4465,"fleet":655},{"id":4505,"title":4506,"description":4507,"docsLink":4508,"tiers":4509},"private-npm-registry","Custom Node-RED Nodes","Create and manage your own private npm registry for Node-RED nodes, so you can share custom nodes across your team and devices without publishing them publicly.","\u002Fdocs\u002Fuser\u002Fcustom-npm-packages\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4511,"title":4512,"description":4513,"docsLink":4514,"changelog":4515,"tiers":4519},"flowfuse-tables","FlowFuse Tables","A managed PostgreSQL database for every application, so you can store and query structured data without standing up and maintaining your own database.","\u002Fdocs\u002Fuser\u002Fff-tables\u002F",[4516],{"url":4517,"release":4518},"\u002Fchangelog\u002F2026\u002F07\u002Fexpert-tables-automation\u002F","2.33",{"edge":655,"hub":655,"fleet":655},{"id":4521,"title":4522,"description":4523,"docsLink":4524,"tiers":4525},"persistent-files","File Storage","Store and retrieve files from your Node-RED flows, with automatic replication and backup across your devices and hosted instances.","\u002Fdocs\u002Finstall\u002Ffile-storage\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4527,"title":4528,"description":4529,"docsLink":4530,"showOnPricing":4465,"tiers":4531},"persistent-context","Persistent Context","In-memory values defined in a Node-RED flow persist across project restarts and upgrades.","\u002Fdocs\u002Fuser\u002Fpersistent-context\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4533,"title":4534,"showOnPricing":4465,"tiers":4535},"static-assets","Static Assets",{"edge":655,"hub":655,"fleet":655},{"id":4537,"title":4538,"description":4539,"docsLink":4540,"tiers":4541},"team-library","Team Library","Set up standard nodes and flows that can be shared with all team members across your organisation.","\u002Fdocs\u002Fuser\u002Fshared-library\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4543,"title":4544,"description":4545,"docsLink":4546,"tiers":4547},"personalised-multi-user-dashboards","Personalised Multi-User Dashboards","Build applications that provide unique data to each logged-in user using personalised multi-user dashboards.","https:\u002F\u002Fdashboard.flowfuse.com\u002Fuser\u002Fmulti-tenancy.html",{"edge":655,"hub":655,"fleet":655},{"id":4549,"title":4550,"description":4551,"changelog":4552,"subfeature":655,"showOnPricing":4465,"tiers":4555},"dashboards-view","Dashboards View","Browse and open every dashboard across your team from a dedicated Dashboards view, at both team and application level, without leaving FlowFuse.",[4553],{"url":4554,"release":4518},"\u002Fchangelog\u002F2026\u002F07\u002Fteam-and-application-dashboards\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4557,"title":4558,"description":4559,"docsLink":4560,"tiers":4561},"blueprints-converge","Blueprints","Ready-made starting points for your apps, from cross-team Converge templates to OT and IT specific blueprints.","\u002Fdocs\u002Fuser\u002Fconcepts\u002F#blueprint",{"edge":655,"hub":655,"fleet":655},{"id":4563,"title":4564,"subfeature":655,"showOnPricing":4465,"tiers":4565},"blueprints-ot-apps","Blueprints - OT APPS",{"edge":655,"hub":4465,"fleet":655},{"id":4567,"title":4568,"subfeature":655,"showOnPricing":4465,"tiers":4569},"blueprints-it-apps","Blueprints - IT APPS",{"edge":4465,"hub":655,"fleet":4465},{"id":4571,"title":4572,"description":4573,"changelog":4574,"showOnPricing":4465,"tiers":4578},"immersive-editor-snapshots","Snapshot Details in Immersive Editor","View and manage snapshot details directly inside the immersive editor without leaving your editing session.",[4575],{"url":4576,"release":4577},"\u002Fchangelog\u002F2026\u002F03\u002Fsnapshot-detail-modal-immersive-editor\u002F","2.29",{"edge":655,"hub":655,"fleet":655},{"id":4580,"title":4581,"description":4582,"changelog":4583,"showOnPricing":4465,"tiers":4586},"immersive-editor-drawer","Customisable Immersive Editor Drawer","Pin, move, resize, or full-screen the immersive editor drawer. Your preferences are remembered between sessions.",[4584],{"url":4585,"release":4475},"\u002Fchangelog\u002F2026\u002F04\u002Fimmersive-editor-drawer\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4588,"title":4589,"description":4590,"changelog":4591,"showOnPricing":4465,"tiers":4594},"embedded-editor-tab-title","Embedded Editor Browser Tab Title","The browser tab title updates to reflect the active Node-RED canvas tab when working in the embedded editor.",[4592],{"url":4593,"release":4577},"\u002Fchangelog\u002F2026\u002F03\u002Fembedded-editor-tab-title\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4596,"title":4597,"features":4598},"deploy","Deploy",[4599,4605,4614,4620,4624,4630,4636,4642,4647],{"id":4600,"title":4601,"description":4602,"docsLink":4603,"tiers":4604},"hosted-instances","Cloud Instances","Run Node-RED instances managed and hosted by FlowFuse.","\u002Fdocs\u002Fuser\u002Fintroduction\u002F#creating-a-node-red-instance",{"edge":655,"hub":655,"fleet":655},{"id":4606,"title":4607,"description":4608,"docsLink":4609,"changelog":4610,"tiers":4613},"edge-devices","Edge Instances","Deploy and mange your Node-RED instances on edge PLCs and gateways, with full visibility and control from the cloud.","\u002Fdocs\u002Fdevice-agent\u002Fintroduction\u002F",[4611],{"url":4612,"release":4455},"\u002Fchangelog\u002F2026\u002F02\u002Fdevice-agent-nodejs-options\u002F",{"edge":655,"hub":4465,"fleet":655},{"id":4615,"title":4616,"description":4617,"docsLink":4618,"tiers":4619},"custom-hostnames","Custom Hostnames","Access your Node-RED application via your own domain name.","\u002Fdocs\u002Fuser\u002Fcustom-hostnames\u002F",{"edge":4465,"hub":655,"fleet":4465},{"id":301,"title":302,"description":4621,"docsLink":4622,"tiers":4623},"Manage and create MQTT clients to transport data for efficient messaging and communication within your applications.","\u002Fdocs\u002Fuser\u002Fteambroker\u002F",{"edge":655,"hub":4465,"fleet":655},{"id":4625,"title":4626,"description":4627,"docsLink":4628,"showOnPricing":4465,"tiers":4629},"project-nodes","Project Nodes aka seamless project comms","FlowFuse Project Nodes enable the passing of data and messages between your Node-RED projects.","\u002Fdocs\u002Fuser\u002Fprojectnodes\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4631,"title":4632,"description":4633,"docsLink":4634,"tiers":4635},"devops-pipelines","DevOps Pipelines","Set up different environments for development, testing, and production Node-RED instances to support a full software delivery lifecycle.","\u002Fdocs\u002Fuser\u002Fdevops-pipelines\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4637,"title":4638,"description":4639,"docsLink":4640,"tiers":4641},"git-integration","Git Integration","Back up your flows to a remote Git repository through a DevOps Pipeline. Supports GitHub and Azure DevOps repositories.","\u002Fdocs\u002Fuser\u002Fdevops-pipelines\u002F#git-repository-stage",{"edge":4465,"hub":655,"fleet":4465},{"id":4643,"title":4644,"description":4645,"docsLink":4640,"subfeature":655,"showOnPricing":4465,"tiers":4646},"git-integration-github","GitHub","Push and pull snapshots to GitHub repositories through DevOps Pipeline Git Stages.",{"edge":4465,"hub":655,"fleet":4465},{"id":4648,"title":4649,"description":4650,"docsLink":4640,"changelog":4651,"subfeature":655,"showOnPricing":4465,"tiers":4654},"git-integration-azure","Azure DevOps","Push and pull snapshots to Azure DevOps repositories through DevOps Pipeline Git Stages.",[4652],{"url":4653,"release":4577},"\u002Fchangelog\u002F2026\u002F03\u002Fazure-dev-ops-gitops\u002F",{"edge":4465,"hub":655,"fleet":4465},{"id":4656,"title":4657,"features":4658},"operate-maintain","Operate & Maintain",[4659,4665,4671,4676,4684,4688,4692,4697,4703,4709,4714,4720,4724,4728],{"id":4660,"title":4661,"description":4662,"docsLink":4663,"tiers":4664},"snapshots","Snapshots & Version History","Automatic snapshots on remote devices and hosted instances, plus a full version history timeline so you can roll back to any prior state.","\u002Fdocs\u002Fuser\u002Fsnapshots\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4666,"title":4667,"description":4668,"docsLink":4669,"subfeature":655,"showOnPricing":4465,"tiers":4670},"auto-snapshot-remote","Auto Snapshot (Remote)","Automatically capture a snapshot every time a remote instance is deployed, so you always have a recoverable history of what was running on each device.","\u002Fdocs\u002Fuser\u002Fsnapshots\u002F#auto-snapshots",{"edge":655,"hub":655,"fleet":655},{"id":4672,"title":4673,"description":4674,"docsLink":4669,"subfeature":655,"showOnPricing":4465,"tiers":4675},"auto-snapshot-hosted","Auto Snapshot (Hosted)","Automatically capture a snapshot every time a hosted instance is deployed, so you always have a recoverable history of what was running.",{"edge":655,"hub":655,"fleet":655},{"id":4677,"title":4678,"description":4679,"changelog":4680,"subfeature":655,"showOnPricing":4465,"tiers":4683},"snapshot-comparison","Snapshot Comparison","Compare two snapshots side-by-side with a navigable diff view. Step through every changed, added, or deleted node and see property and code diffs.",[4681],{"url":4682,"release":4577},"\u002Fchangelog\u002F2026\u002F04\u002Fsnapshot-diff-viewer\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4685,"title":4686,"subfeature":655,"showOnPricing":4465,"tiers":4687},"version-history-timeline","Version History Timeline",{"edge":655,"hub":655,"fleet":655},{"id":4689,"title":4690,"tiers":4691},"unlimited-workflow-executions","Unlimited Workflow Executions",{"edge":655,"hub":655,"fleet":655},{"id":4693,"title":4694,"description":4695,"tiers":4696},"device-fleet-updates","Device Fleet Updates","Connect to edge devices to quickly assess and update logic. Debug one device and roll out improvements to your fleet in minutes, securely without requiring full device access for your whole organisation.",{"edge":655,"hub":4465,"fleet":655},{"id":4698,"title":4699,"description":4700,"docsLink":4701,"tiers":4702},"device-group-management","Device Group Management","Logically group devices assigned to an application and integrate device groups into your DevOps Pipeline for coordinated fleet updates.","\u002Fdocs\u002Fuser\u002Fdevice-groups\u002F",{"edge":655,"hub":4465,"fleet":655},{"id":4704,"title":4705,"description":4706,"docsLink":4707,"tiers":4708},"high-availability","High Availability","Leverage horizontal scaling for reliable and scalable processing of your data through Node-RED.","\u002Fdocs\u002Fuser\u002Fhigh-availability\u002F",{"edge":4465,"hub":655,"fleet":4465},{"id":4710,"title":4711,"description":4712,"tiers":4713},"performance-monitoring","Performance Monitoring & Alerts","Track CPU, memory, and event loop performance across your instances and devices, with email alerts when something needs your attention.",{"edge":655,"hub":655,"fleet":655},{"id":4715,"title":4716,"description":4717,"docsLink":4718,"subfeature":655,"showOnPricing":4465,"tiers":4719},"instance-monitoring","Instance Monitoring","Enable alerts to be sent via email when your Node-RED instances encounter issues.","\u002Fdocs\u002Fuser\u002Finstance-settings\u002F#alerts",{"edge":655,"hub":655,"fleet":655},{"id":4721,"title":4722,"subfeature":655,"showOnPricing":4465,"tiers":4723},"email-alerts","Email Alerts",{"edge":655,"hub":655,"fleet":655},{"id":4725,"title":4726,"showOnPricing":4465,"tiers":4727},"api-debug-length-limit","API\u002FDebug Length Limit",{"edge":655,"hub":655,"fleet":655},{"id":4729,"title":4730,"tiers":4731},"protected-instances","Protected Instances",{"edge":4465,"hub":655,"fleet":4465},{"id":4733,"title":4734,"features":4735},"govern-secure","Govern and Secure",[4736,4745,4751,4757,4762,4768,4774,4780,4788,4794,4800,4806],{"id":4737,"title":4738,"description":4739,"docsLink":4740,"changelog":4741,"tiers":4744},"single-sign-on","Single Sign-On (SSO)","Configure FlowFuse to work with your own SSO provider, allowing users to access FlowFuse with a single set of login credentials.","\u002Fdocs\u002Fadmin\u002Fsso\u002F",[4742],{"url":4743,"release":4518},"\u002Fchangelog\u002F2026\u002F07\u002Fapplication-sso-groups\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4746,"title":4747,"description":4748,"docsLink":4749,"tiers":4750},"two-factor-authentication","Two-Factor Authentication","Two-factor authentication adds an extra layer of security to your FlowFuse account.","\u002Fdocs\u002Fuser\u002Fuser-settings\u002F#two-factor-authentication",{"edge":655,"hub":655,"fleet":655},{"id":4752,"title":4753,"description":4754,"docsLink":4755,"tiers":4756},"certified-nodes-it","Certified Nodes - IT","IT certified node bundle includes: Redis, MQTT, HTTP Request, AI nodes (Gemini, Claude, ChatGPT, Ollama), MCP Server","\u002Fblog\u002F2025\u002F07\u002Fcertified-nodes-v2\u002F",{"edge":4465,"hub":655,"fleet":655},{"id":4758,"title":4759,"description":4760,"tiers":4761},"certified-nodes-ot","Certified OT Connections - OPC-UA, Modbus, etc.","OT certified node bundle includes: OPC-UA, Modbus TCP & RTU, RTSP, EtherNet\u002FIP, AI nodes (Gemini, Claude, ChatGPT, Ollama), MCP Server",{"edge":655,"hub":4465,"fleet":4465},{"id":4763,"title":4764,"description":4765,"docsLink":4766,"tiers":4767},"audit-log","Audit Log","Keep track of everything going on in your Node-RED instances and FlowFuse. Audit Logs provide details on what actions have taken place, when they happened, and who did them.","\u002Fdocs\u002Fuser\u002Flogs\u002F#audit-log",{"edge":655,"hub":655,"fleet":655},{"id":4769,"title":4770,"description":4771,"docsLink":4772,"tiers":4773},"role-based-access-control","Role-Based Access Control","Control who can do what at both the team and application level, from viewers to admins.","\u002Fdocs\u002Fuser\u002Frole-based-access-control\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4775,"title":4776,"description":4777,"docsLink":4778,"subfeature":655,"showOnPricing":4465,"tiers":4779},"application-level-rbac","Application-Level RBAC","Fine-grained access control per application, allowing team members to have different permission levels across different applications without requiring separate teams.","\u002Fdocs\u002Fuser\u002Frole-based-access-control\u002F#application-level-rbac",{"edge":655,"hub":655,"fleet":655},{"id":4781,"title":4782,"description":4783,"changelog":4784,"subfeature":655,"showOnPricing":4465,"tiers":4787},"scoped-personal-access-tokens","Scoped Personal Access Tokens","Restrict a Personal Access Token to specific teams, limit it to read-only operations, or control whether it carries admin privileges.",[4785],{"url":4786,"release":4518},"\u002Fchangelog\u002F2026\u002F07\u002Fscoped-pats\u002F",{"edge":655,"hub":655,"fleet":655},{"id":4789,"title":4790,"description":4791,"docsLink":4792,"showOnPricing":4465,"tiers":4793},"team-members","Team Members","Invite multiple team members to collaborate on the same Node-RED flows.","\u002Fdocs\u002Fuser\u002Fteam\u002F#teams",{"edge":655,"hub":655,"fleet":655},{"id":4795,"title":4796,"description":4797,"docsLink":4798,"showOnPricing":4465,"tiers":4799},"endpoint-security","Endpoint Security","Secure HTTP endpoints for hosted Node-RED instances using FlowFuse credentials.","\u002Fdocs\u002Fuser\u002Finstance-settings\u002F#security",{"edge":655,"hub":655,"fleet":655},{"id":4801,"title":4802,"description":4803,"docsLink":4804,"tiers":4805},"baa-for-hipaa","BAA for HIPAA","FlowFuse can sign a Business Associate Agreement to ensure proper safeguarding of protected health information handled on your behalf.","\u002Fhandbook\u002Fsales\u002Fsubscription-agreement-1.5\u002F",{"edge":4465,"hub":655,"fleet":4465},{"id":4807,"title":4808,"description":4809,"tiers":4810},"sbom","Software Bill of Materials","A complete list of all software components used in your Node-RED instances and hosted applications, including version numbers and license information.",{"edge":4465,"hub":655,"fleet":4465},{"id":4812,"title":4813,"features":4814},"support","Support",[4815,4820,4824],{"id":4816,"title":4817,"docsLink":4818,"tiers":4819},"installation-support","Installation Support","\u002Fdocs\u002Finstall\u002Fintroduction\u002F#do-you-need-help-installation-service",{"edge":655,"hub":655,"fleet":655},{"id":4821,"title":4822,"showOnPricing":4465,"tiers":4823},"live-chat-support","Live Chat Support",{"edge":655,"hub":655,"fleet":655},{"id":4825,"title":4826,"docsLink":4827,"tiers":4828},"enterprise-support","Enterprise Support","\u002Fdocs\u002Fpremium-support\u002F",{"edge":655,"hub":655,"fleet":655},"feature-catalog","0AnkhTIPCECCwP9NmbTWP5HvRYx4FTSao4lG93-HaWM",[4832,4835,4838,4841,4844,4847,4849,4852,4855,4858,4861,4863,4866,4869,4872,4875,4878,4881,4884,4887,4890,4893,4896,4899,4902,4905,4908,4911,4914,4917,4920,4923,4926,4929,4932,4935,4938,4941,4944,4947,4950,4953,4956,4959,4962,4965,4968,4971,4974,4976,4979,4982,4985,4988,4991,4994,4997,4999,5002,5005,5008,5011,5014,5016,5019,5022,5025,5028,5031,5034,5037,5040,5043,5046,5048,5051,5054,5057,5060,5063,5066,5069,5072,5075,5078,5081,5084,5087,5090,5092,5095,5098,5101,5104,5107,5110,5113,5116,5119,5122,5125,5128,5131,5134,5137,5140,5143,5146,5148,5151,5154,5157,5160,5163,5166,5169,5171,5174,5177,5180,5183,5186,5189,5192,5195,5198,5201,5204,5207,5210,5213,5216,5219,5222,5225,5228,5231,5234,5237,5240,5243,5246,5249,5252,5255,5258,5261,5264,5267,5270,5273,5276,5279,5282,5285,5288,5291,5294,5297,5300,5303,5306,5309,5312,5315,5318,5321,5323,5326,5329,5332,5335,5338,5341,5344,5347,5350,5353,5356,5359,5362],{"path":4833,"title":4834},"\u002Fchangelog\u002F2023\u002F09\u002Fcustom-node-support","Custom Node Support",{"path":4836,"title":4837},"\u002Fchangelog\u002F2023\u002F09\u002Fdevops-actions","DevOps Pipeline with action selection",{"path":4839,"title":4840},"\u002Fchangelog\u002F2023\u002F09\u002Fintroduction-enterprise-tier","Introducing the Enterprise Tier",{"path":4842,"title":4843},"\u002Fchangelog\u002F2023\u002F09\u002Fpipeline-api","API Endpoint for DevOps Pipeline",{"path":4845,"title":4846},"\u002Fchangelog\u002F2023\u002F09\u002Fsnapshots-devices","Usability improvements to Device Management",{"path":4848,"title":4558},"\u002Fchangelog\u002F2023\u002F10\u002Fblueprints",{"path":4850,"title":4851},"\u002Fchangelog\u002F2023\u002F10\u002Fcertified-nodes","Certified Nodes",{"path":4853,"title":4854},"\u002Fchangelog\u002F2023\u002F10\u002Fdevice-snapshot-selection","Enhanced Snapshot Selection",{"path":4856,"title":4857},"\u002Fchangelog\u002F2023\u002F10\u002Fpath-bug-fix","Device Agent path bug fix",{"path":4859,"title":4860},"\u002Fchangelog\u002F2023\u002F10\u002Fresource-alerts","Resource Monitoring in Audit Log",{"path":4862,"title":4747},"\u002Fchangelog\u002F2023\u002F11\u002F2fa",{"path":4864,"title":4865},"\u002Fchangelog\u002F2023\u002F11\u002Fdefault-editor","Device Editor enabled by default",{"path":4867,"title":4868},"\u002Fchangelog\u002F2023\u002F11\u002Fdevices-in-pipelines","Devices in DevOps Pipelines",{"path":4870,"title":4871},"\u002Fchangelog\u002F2023\u002F11\u002Fproject-nodes-devices","Project Nodes for Devices",{"path":4873,"title":4874},"\u002Fchangelog\u002F2023\u002F12\u002Fbilling","No Credit Card required for billing",{"path":4876,"title":4877},"\u002Fchangelog\u002F2023\u002F12\u002Fblueprint-selection","Blueprint Selection Update",{"path":4879,"title":4880},"\u002Fchangelog\u002F2023\u002F12\u002Fdevice-groups","Device Groups",{"path":4882,"title":4883},"\u002Fchangelog\u002F2023\u002F12\u002Femail-alerting-node-red-crash","Email Alerts for Audit Log Events",{"path":4885,"title":4886},"\u002Fchangelog\u002F2023\u002F12\u002Fnode-red-updated","Node-RED 3.1.3 now available",{"path":4888,"title":4889},"\u002Fchangelog\u002F2024\u002F01\u002Fdevice-audit-log","Introducing the Device Auditlog Feature",{"path":4891,"title":4892},"\u002Fchangelog\u002F2024\u002F01\u002Fdevice-groups-snapshot","Device Groups - Automatic snapshot assignment",{"path":4894,"title":4895},"\u002Fchangelog\u002F2024\u002F01\u002Ffleet-mode","Renaming \"Default Device Mode\" to \"Fleet Mode\"",{"path":4897,"title":4898},"\u002Fchangelog\u002F2024\u002F01\u002Fhelm-v2","Helm Chart v2.0",{"path":4900,"title":4901},"\u002Fchangelog\u002F2024\u002F01\u002Fnew-blueprints","New Blueprints added",{"path":4903,"title":4904},"\u002Fchangelog\u002F2024\u002F01\u002Fsecurity-updates","Security Updates",{"path":4906,"title":4907},"\u002Fchangelog\u002F2024\u002F01\u002Fsso-team-membership","Managing Team Membership via SSO",{"path":4909,"title":4910},"\u002Fchangelog\u002F2024\u002F01\u002Fstreamlined-device-assignment","Streamlined Device assignment",{"path":4912,"title":4913},"\u002Fchangelog\u002F2024\u002F02\u002Fdevice-auto-snapshot","Device Auto Snapshots",{"path":4915,"title":4916},"\u002Fchangelog\u002F2024\u002F02\u002Fdevice-instance-audit-logs","Device Instance Audit Logging",{"path":4918,"title":4919},"\u002Fchangelog\u002F2024\u002F02\u002Fdevice-onboarding-improvements","Device Onboarding Improvements",{"path":4921,"title":4922},"\u002Fchangelog\u002F2024\u002F02\u002Fdevice-pricing-change","Pricing change for Devices",{"path":4924,"title":4925},"\u002Fchangelog\u002F2024\u002F02\u002Finstance-auto-snapshots","Instance Auto Snapshots",{"path":4927,"title":4928},"\u002Fchangelog\u002F2024\u002F02\u002Fpostgresql-upgrade","PostgreSQL Version update",{"path":4930,"title":4931},"\u002Fchangelog\u002F2024\u002F03\u002Fbearer-token-authentication","Bearer Token Authentication for Node-RED Instances",{"path":4933,"title":4934},"\u002Fchangelog\u002F2024\u002F03\u002Finstance-protection-mode","Instance Protection Mode",{"path":4936,"title":4937},"\u002Fchangelog\u002F2024\u002F03\u002Flimits-debug-payload","Configure HTTP Payload and Debug Message size",{"path":4939,"title":4940},"\u002Fchangelog\u002F2024\u002F03\u002Frestart-devices-remotly","Remote Device Restart",{"path":4942,"title":4943},"\u002Fchangelog\u002F2024\u002F04\u002Fcustom-nodes-on-devices","Custom Nodes Support on Devices",{"path":4945,"title":4946},"\u002Fchangelog\u002F2024\u002F04\u002Fdevice-auto-snapshot","Direct Dashboard Access",{"path":4948,"title":4949},"\u002Fchangelog\u002F2024\u002F04\u002Fimproving-device-groups","Improving Device Groups",{"path":4951,"title":4952},"\u002Fchangelog\u002F2024\u002F04\u002Fpricing-change","Pricing change Enterprise & Teams Tier",{"path":4954,"title":4955},"\u002Fchangelog\u002F2024\u002F04\u002Ftougher-rate-limiting","Tougher Rate Limiting on Public Routes",{"path":4957,"title":4958},"\u002Fchangelog\u002F2024\u002F05\u002Finstance-healthcheck","Customizing instance health-check settings",{"path":4960,"title":4961},"\u002Fchangelog\u002F2024\u002F05\u002Flibrary-blueprints","Blueprints added to Library",{"path":4963,"title":4964},"\u002Fchangelog\u002F2024\u002F05\u002Flibrary-flowviewer","Team Library - Flow Viewer",{"path":4966,"title":4967},"\u002Fchangelog\u002F2024\u002F05\u002Fmanaging-node-red-version-on-devices","Managing Node-RED versions on Devices",{"path":4969,"title":4970},"\u002Fchangelog\u002F2024\u002F05\u002Fsnapshot-improvements","Snapshot Improvements",{"path":4972,"title":4973},"\u002Fchangelog\u002F2024\u002F05\u002Fsnapshot-improvements-pt3","Snapshot Upload",{"path":4975,"title":4973},"\u002Fchangelog\u002F2024\u002F05\u002Fsnapshot-upload",{"path":4977,"title":4978},"\u002Fchangelog\u002F2024\u002F06\u002Fdevice-agent-proxy-support","Running the Device Agent behind an HTTP proxy",{"path":4980,"title":4981},"\u002Fchangelog\u002F2024\u002F06\u002Flibrary-blueprints","Custom hostnames for your instances",{"path":4983,"title":4984},"\u002Fchangelog\u002F2024\u002F06\u002Fmultiline-env-vars","Multi-line Environment Variables",{"path":4986,"title":4987},"\u002Fchangelog\u002F2024\u002F06\u002Fsnapshot-flow-compare","Compare Snapshots flows",{"path":4989,"title":4990},"\u002Fchangelog\u002F2024\u002F07\u002Fapplications-search","Applications Search",{"path":4992,"title":4993},"\u002Fchangelog\u002F2024\u002F07\u002Fdevice-group-clear-snapshot","Device Groups Snapshots",{"path":4995,"title":4996},"\u002Fchangelog\u002F2024\u002F07\u002Fdevice-management-bulk-delete","Managing devices",{"path":4998,"title":4996},"\u002Fchangelog\u002F2024\u002F07\u002Fdevice-management-bulk-move",{"path":5000,"title":5001},"\u002Fchangelog\u002F2024\u002F07\u002Fedit-snapshots","Edit Snapshots",{"path":5003,"title":5004},"\u002Fchangelog\u002F2024\u002F07\u002Fflowfuse-assistant","The FlowFuse Expert",{"path":5006,"title":5007},"\u002Fchangelog\u002F2024\u002F07\u002Fflowfuse-assistant-json","FlowFuse Expert Writes JSON",{"path":5009,"title":5010},"\u002Fchangelog\u002F2024\u002F07\u002Fimmersive-editor","Immersive Editor Experience",{"path":5012,"title":5013},"\u002Fchangelog\u002F2024\u002F07\u002Fnotifications-inbox","Notifications Inbox",{"path":5015,"title":5013},"\u002Fchangelog\u002F2024\u002F07\u002Fnotifications-update",{"path":5017,"title":5018},"\u002Fchangelog\u002F2024\u002F07\u002Fpersistent-storage","Persistent Storage on FlowFuse Cloud",{"path":5020,"title":5021},"\u002Fchangelog\u002F2024\u002F07\u002Fsso","Single Sign On Updates",{"path":5023,"title":5024},"\u002Fchangelog\u002F2024\u002F08\u002Fbill-of-materials","Bill of Materials",{"path":5026,"title":5027},"\u002Fchangelog\u002F2024\u002F08\u002Fenterprise-license-update","Enforcing Enterprise Restrictions",{"path":5029,"title":5030},"\u002Fchangelog\u002F2024\u002F08\u002Fldap-sso-groups","LDAP Single Sign On Updates",{"path":5032,"title":5033},"\u002Fchangelog\u002F2024\u002F08\u002Fstatic-file-service-navigation-visibility","Static File Service Navigation and Visibility",{"path":5035,"title":5036},"\u002Fchangelog\u002F2024\u002F10\u002Fdevice-group-env-vars","Environment Variables for your Device Groups",{"path":5038,"title":5039},"\u002Fchangelog\u002F2024\u002F10\u002Fmqtt-service","MQTT Broker Service",{"path":5041,"title":5042},"\u002Fchangelog\u002F2024\u002F10\u002Fnotifications-bulk-actions","Managing Notifications",{"path":5044,"title":5045},"\u002Fchangelog\u002F2024\u002F10\u002Fsnapshot-download-upload-options","Snapshot Upload and Download Improvements",{"path":5047,"title":4686},"\u002Fchangelog\u002F2024\u002F10\u002Fversion-history-timeline",{"path":5049,"title":5050},"\u002Fchangelog\u002F2024\u002F11\u002Faudit-log-hierarchy","Audit logs show hierarchical events",{"path":5052,"title":5053},"\u002Fchangelog\u002F2024\u002F11\u002Fdevice-agent-release","Device Agent 3.0 released",{"path":5055,"title":5056},"\u002Fchangelog\u002F2024\u002F11\u002Fmqtt-topic-hierarchy","MQTT Topic Hierarchy view",{"path":5058,"title":5059},"\u002Fchangelog\u002F2024\u002F11\u002Fteam-search","Team-wide search",{"path":5061,"title":5062},"\u002Fchangelog\u002F2024\u002F12\u002Fdashboad-iframe","Allow Dashboards to be embedded in iFrames",{"path":5064,"title":5065},"\u002Fchangelog\u002F2024\u002F12\u002Fdevice-editor-cache","Device Editor Access Speed Up",{"path":5067,"title":5068},"\u002Fchangelog\u002F2024\u002F12\u002Fteam-bom-timeline","Team BOM and Pipeline Views",{"path":5070,"title":5071},"\u002Fchangelog\u002F2025\u002F01\u002Ffree-tier-onboarding","New Onboarding Tour for Free Tier Users",{"path":5073,"title":5074},"\u002Fchangelog\u002F2025\u002F01\u002Fhidden-env-vars","Hidden Environment Variables",{"path":5076,"title":5077},"\u002Fchangelog\u002F2025\u002F01\u002Fimproved-diagnostics","Improved Diagnostics",{"path":5079,"title":5080},"\u002Fchangelog\u002F2025\u002F01\u002Fteam-level-groups","Team level view of Groups",{"path":5082,"title":5083},"\u002Fchangelog\u002F2025\u002F02\u002Fadditional-device-version-history-events","New Remote Instances Version History Events",{"path":5085,"title":5086},"\u002Fchangelog\u002F2025\u002F02\u002Fbroker-error-feedback","Improved Broker Connection Feedback",{"path":5088,"title":5089},"\u002Fchangelog\u002F2025\u002F02\u002Fdevice-agent-updates","FlowFuse User Authentication on Remote Instances",{"path":5091,"title":4686},"\u002Fchangelog\u002F2025\u002F02\u002Fdevice-version-history-timeline",{"path":5093,"title":5094},"\u002Fchangelog\u002F2025\u002F02\u002Fexternal-brokers","External MQTT Brokers",{"path":5096,"title":5097},"\u002Fchangelog\u002F2025\u002F02\u002Fmqtt-schema-suggestions","MQTT Smart Schema Suggestions",{"path":5099,"title":5100},"\u002Fchangelog\u002F2025\u002F02\u002Fresend-and-extend-team-invitation-expiration","Re-send Team Invitations",{"path":5102,"title":5103},"\u002Fchangelog\u002F2025\u002F02\u002Fschema-docs","Personalized Schema Documentation",{"path":5105,"title":5106},"\u002Fchangelog\u002F2025\u002F02\u002Ftopic-hierarchy-search","Search & Filter for Topic Hierarchy List",{"path":5108,"title":5109},"\u002Fchangelog\u002F2025\u002F03\u002Fcontainer-tags","Changes to tags for flowfuse\u002Fnode-red",{"path":5111,"title":5112},"\u002Fchangelog\u002F2025\u002F03\u002Fdevice-groups","Multiple Device Groups in a pipeline",{"path":5114,"title":5115},"\u002Fchangelog\u002F2025\u002F03\u002Fdevice-local-login","Local Login for Remote Instances",{"path":5117,"title":5118},"\u002Fchangelog\u002F2025\u002F03\u002Ffree-tier","Free Tier now more accessible to all",{"path":5120,"title":5121},"\u002Fchangelog\u002F2025\u002F03\u002Fresource-notifications","Resource Alerts",{"path":5123,"title":5124},"\u002Fchangelog\u002F2025\u002F03\u002Fsnapshot-filter","Filtering Snapshots",{"path":5126,"title":5127},"\u002Fchangelog\u002F2025\u002F03\u002Fteam-npm-registry","NPM Package Hosting",{"path":5129,"title":5130},"\u002Fchangelog\u002F2025\u002F03\u002Ftopic-deletion","MQTT Topic Management",{"path":5132,"title":5133},"\u002Fchangelog\u002F2025\u002F04\u002Fdevice-provisioning","Remote Instance Provisioning",{"path":5135,"title":5136},"\u002Fchangelog\u002F2025\u002F04\u002Fgit-integration","Git Integration with Pipelines",{"path":5138,"title":5139},"\u002Fchangelog\u002F2025\u002F04\u002Finstance-log-browsing","Better Node-RED log handling",{"path":5141,"title":5142},"\u002Fchangelog\u002F2025\u002F05\u002Fimport-node-red-flows","Import Node-RED Flows During Remote Instance Setup",{"path":5144,"title":5145},"\u002Fchangelog\u002F2025\u002F06\u002Fflowfuse-assistant","FlowFuse Expert just got smarter",{"path":5147,"title":5145},"\u002Fchangelog\u002F2025\u002F06\u002Fflowfuse-assistant-2",{"path":5149,"title":5150},"\u002Fchangelog\u002F2025\u002F06\u002Fgit-integration","Pulling snapshots from Git with Pipelines",{"path":5152,"title":5153},"\u002Fchangelog\u002F2025\u002F06\u002Finstance-performance-memory","Memory Metrics in Instance Performance View",{"path":5155,"title":5156},"\u002Fchangelog\u002F2025\u002F06\u002Fnew-home-page","Introducing the New Home Page Experience",{"path":5158,"title":5159},"\u002Fchangelog\u002F2025\u002F06\u002Fteam-performance","Team Performance Feature",{"path":5161,"title":5162},"\u002Fchangelog\u002F2025\u002F06\u002Fteam-performance-view","Instance Performance View",{"path":5164,"title":5165},"\u002Fchangelog\u002F2025\u002F06\u002Fui-refresh","Navigation UI Refresh",{"path":5167,"title":5168},"\u002Fchangelog\u002F2025\u002F07\u002Fbrowse-node-red-flows","Browse for Node-RED Flows During Remote Instance Setup",{"path":5170,"title":4512},"\u002Fchangelog\u002F2025\u002F07\u002Fflowfuse-tables",{"path":5172,"title":5173},"\u002Fchangelog\u002F2025\u002F07\u002Fimport-blueprints","Import blueprints directly into your existing instances",{"path":5175,"title":5176},"\u002Fchangelog\u002F2025\u002F07\u002Fsimplified-applications-overview","Simplified Applications Page with Summary Tiles",{"path":5178,"title":5179},"\u002Fchangelog\u002F2025\u002F07\u002Fsmart-suggestions","Smart Suggestions",{"path":5181,"title":5182},"\u002Fchangelog\u002F2025\u002F07\u002Fteam-to-pro-plan-rename","Team Plan Renamed to Pro Plan",{"path":5184,"title":5185},"\u002Fchangelog\u002F2025\u002F08\u002Fai-generated-snapshot-descriptions-hosted","Generate snapshot descriptions with AI",{"path":5187,"title":5188},"\u002Fchangelog\u002F2025\u002F08\u002Fai-generated-snapshot-descriptions-remote","AI Snapshot Descriptions Now Work with Remote Instances",{"path":5190,"title":5191},"\u002Fchangelog\u002F2025\u002F08\u002Fdevice-performance","FlowFuse Remote Instance Performance Data",{"path":5193,"title":5194},"\u002Fchangelog\u002F2025\u002F08\u002Fdirect-sso","Direct SSO Login",{"path":5196,"title":5197},"\u002Fchangelog\u002F2025\u002F08\u002Fflowfuse-assistant","FlowFuse Expert documents your flows",{"path":5199,"title":5200},"\u002Fchangelog\u002F2025\u002F08\u002Fflowfuse-mqtt","FlowFuse MQTT",{"path":5202,"title":5203},"\u002Fchangelog\u002F2025\u002F08\u002Fhttp-cors","Configure HTTP CORS",{"path":5205,"title":5206},"\u002Fchangelog\u002F2025\u002F08\u002Fsubflow-export","Export SubFlow as Node-RED module",{"path":5208,"title":5209},"\u002Fchangelog\u002F2025\u002F08\u002Ftables-assistant","FlowFuse Tables with a little help from the Assistant",{"path":5211,"title":5212},"\u002Fchangelog\u002F2025\u002F09\u002Fexpose-saml-groups-to-dashboard","Allow SSO groups to be shared with the Node-RED Dashboard",{"path":5214,"title":5215},"\u002Fchangelog\u002F2025\u002F09\u002Finline-assist","FlowFuse Expert can help you write code",{"path":5217,"title":5218},"\u002Fchangelog\u002F2025\u002F09\u002Fretiring-flowforge-device-agent","No more @flowforge\u002Fflowforge-device-agent releases",{"path":5220,"title":5221},"\u002Fchangelog\u002F2025\u002F09\u002Frevised-instance-snapshot-ui","Streamlined Snapshot Management",{"path":5223,"title":5224},"\u002Fchangelog\u002F2025\u002F09\u002Fteam-broker-async-api","Capture Async API data from FlowFuse Team Broker",{"path":5226,"title":5227},"\u002Fchangelog\u002F2025\u002F10\u002Fapplication-level-rbac","Application-level access control for Enterprise teams",{"path":5229,"title":5230},"\u002Fchangelog\u002F2025\u002F10\u002Fbulk-device-group-assignment","Easier Access to Bulk Device Group Management",{"path":5232,"title":5233},"\u002Fchangelog\u002F2025\u002F10\u002Fduplicate-instances-across-applications","Duplicate Instances Across Different Applications",{"path":5235,"title":5236},"\u002Fchangelog\u002F2025\u002F10\u002Fimport-flows-on-instance-creation","Import flows during instance creation",{"path":5238,"title":5239},"\u002Fchangelog\u002F2025\u002F10\u002Fmcp-nodes","FlowFuse MCP Server Nodes",{"path":5241,"title":5242},"\u002Fchangelog\u002F2025\u002F10\u002Fonnx-nodes","FlowFuse AI Nodes",{"path":5244,"title":5245},"\u002Fchangelog\u002F2025\u002F10\u002Fsettings-page-device-group-management","Device Group Management from Settings Page",{"path":5247,"title":5248},"\u002Fchangelog\u002F2025\u002F11\u002Fff-expert-update","FlowFuse Expert Update",{"path":5250,"title":5251},"\u002Fchangelog\u002F2025\u002F11\u002Fminimum-nodejs-version","Node.js v20 Minimum Version Requirement",{"path":5253,"title":5254},"\u002Fchangelog\u002F2025\u002F11\u002Fsso-session","SSO control over Session life",{"path":5256,"title":5257},"\u002Fchangelog\u002F2025\u002F12\u002Fff-expert-mcp-insights","FlowFuse Expert: Now with MCP-Powered Insights",{"path":5259,"title":5260},"\u002Fchangelog\u002F2025\u002F12\u002Fscheduled-maintenance","Scheduled Maintenance Mode",{"path":5262,"title":5263},"\u002Fchangelog\u002F2026\u002F01\u002Fdevice-agent-containers","Device Agent Docker Containers updated",{"path":5265,"title":5266},"\u002Fchangelog\u002F2026\u002F01\u002Fff-expert-manage-palette","FlowFuse Expert: Enhanced Palette Integration",{"path":5268,"title":5269},"\u002Fchangelog\u002F2026\u002F01\u002Fff-expert-nr-actions","FlowFuse Expert: Integration with Node-RED",{"path":5271,"title":5272},"\u002Fchangelog\u002F2026\u002F01\u002Fff-expert-palette-queries","FlowFuse Expert: Palette Queries",{"path":5274,"title":5275},"\u002Fchangelog\u002F2026\u002F01\u002Fff-expert-select-flows","FlowFuse Expert: Ask about your flows",{"path":5277,"title":5278},"\u002Fchangelog\u002F2026\u002F01\u002Fmcp-rbacs","FlowFuse Expert: MCP-Powered Insights with RBACs",{"path":5280,"title":5281},"\u002Fchangelog\u002F2026\u002F01\u002Fmcp-security","FlowFuse Expert: MCP-Powered Insights",{"path":5283,"title":5284},"\u002Fchangelog\u002F2026\u002F02\u002Fdevice-agent-nodejs-options","Set Node.js Options for Remote Instances",{"path":5286,"title":5287},"\u002Fchangelog\u002F2026\u002F02\u002Fff-expert-debug-log-context","FlowFuse Expert: Helping you make sense of your debug log",{"path":5289,"title":5290},"\u002Fchangelog\u002F2026\u002F02\u002Fff-expert-update-banner","FlowFuse Expert: Never Miss an Update",{"path":5292,"title":5293},"\u002Fchangelog\u002F2026\u002F02\u002Fha-instance-rolling-restart","HA Hosted Instance Rolling Restart",{"path":5295,"title":5296},"\u002Fchangelog\u002F2026\u002F02\u002Fremote-instances-immersive-editor","Immersive Mode for Remote Instances",{"path":5298,"title":5299},"\u002Fchangelog\u002F2026\u002F02\u002Frestoring-snapshots-to-remote-instances","Restoring snapshots to developer-mode Remote Instances",{"path":5301,"title":5302},"\u002Fchangelog\u002F2026\u002F03\u002Fazure-dev-ops-gitops","Azure DevOps Pipeline support",{"path":5304,"title":5305},"\u002Fchangelog\u002F2026\u002F03\u002Fdeveloper-mode-in-immersive-editor","Developer Mode Now Accessible from the Immersive Editor",{"path":5307,"title":5308},"\u002Fchangelog\u002F2026\u002F03\u002Fembedded-editor-tab-title","Embedded Editor Now Shows Active Tab",{"path":5310,"title":5311},"\u002Fchangelog\u002F2026\u002F03\u002Fmarch-scheduled-maintenance","Updated: Upcoming Scheduled Server Maintenance on March 28th, 2026",{"path":5313,"title":5314},"\u002Fchangelog\u002F2026\u002F03\u002Fsnapshot-detail-modal-immersive-editor","View Snapshot Details in the Immersive Editor",{"path":5316,"title":5317},"\u002Fchangelog\u002F2026\u002F04\u002Fexpert-action-links","FlowFuse expert action links",{"path":5319,"title":5320},"\u002Fchangelog\u002F2026\u002F04\u002Fhosted-instance-url-env-var","Hosted Instances know their own URL",{"path":5322,"title":4581},"\u002Fchangelog\u002F2026\u002F04\u002Fimmersive-editor-drawer",{"path":5324,"title":5325},"\u002Fchangelog\u002F2026\u002F04\u002Fsnapshot-diff-viewer","Richer snapshot comparison view",{"path":5327,"title":5328},"\u002Fchangelog\u002F2026\u002F05\u002Fai-opt-out","AI Feature Opt-Out for Teams",{"path":5330,"title":5331},"\u002Fchangelog\u002F2026\u002F05\u002Fexpert-application-building","FlowFuse Expert Builds Your Application",{"path":5333,"title":5334},"\u002Fchangelog\u002F2026\u002F05\u002Fsingle-sso-provider","Single SSO provider for all users",{"path":5336,"title":5337},"\u002Fchangelog\u002F2026\u002F06\u002Fdark-mode","Dark Mode",{"path":5339,"title":5340},"\u002Fchangelog\u002F2026\u002F06\u002Fdevice-agent-v4-released","Device Agent v4 released",{"path":5342,"title":5343},"\u002Fchangelog\u002F2026\u002F06\u002Fgeneric-git-server-support","Connect Pipelines to Any Git Server",{"path":5345,"title":5346},"\u002Fchangelog\u002F2026\u002F06\u002Fjuly-scheduled-maintenance","Updated: Upcoming Scheduled Server Maintenance on July 11th, 2026",{"path":5348,"title":5349},"\u002Fchangelog\u002F2026\u002F07\u002Fapplication-sso-groups","Application Roles from SSO Groups",{"path":5351,"title":5352},"\u002Fchangelog\u002F2026\u002F07\u002Fdevice-agent-onboarding","Getting started with FlowFuse straight from the terminal",{"path":5354,"title":5355},"\u002Fchangelog\u002F2026\u002F07\u002Fexpert-enhancements","FlowFuse Expert Enhancements",{"path":5357,"title":5358},"\u002Fchangelog\u002F2026\u002F07\u002Fexpert-tables-automation","FlowFuse Expert Can Now Work With Your Tables",{"path":5360,"title":5361},"\u002Fchangelog\u002F2026\u002F07\u002Fscoped-pats","Team Scoped Personal Access Tokens",{"path":5363,"title":5364},"\u002Fchangelog\u002F2026\u002F07\u002Fteam-and-application-dashboards","A Dedicated Home for Your Dashboards",[],1786105898866]