🏭 MotorSafe: Tri-Modal Edge-AI Predictive Maintenance & 3D Digital Twin

Local-first anomaly isolation and 3D telemetry spatial mapping for Industry 4.0.

MotorSafe is an enterprise-grade, edge-native predictive maintenance ecosystem designed to prevent catastrophic industrial motor failures. By synchronizing and fusing three high-frequency telemetry streams—Voltage (power quality), Current (torque load), and Vibration (mechanical health)—the local unsupervised Machine Learning model (Isolation Forest) cross-verifies anomalies in real-time, eliminating false positives and predicting mechanical degradation days in advance.

---

🏗️ System & Network Architecture

MotorSafe decouples data acquisition, ingestion, and 3D spatial telemetry visualization into a robust wireless Publish/Subscribe topology:

flowchart TD
    subgraph Edge Nodes [Industrial Hardware Nodes]
        A["ACS712 (Current Sensor)"] & B["ZMPT101B (Voltage Sensor)"] & C["MPU6050 (Vibration Accel)"] -->|Analog / I2C| D["ESP8266 Gateway / Publisher"]
    end
    
    D -->|MQTT Over WiFi JSON| E["Mosquitto MQTT Broker (factory/+/telemetry)"]
    
    subgraph Diagnostic Core [Central Analytical Server]
        E -->|Instant Telemetry Sync| F["Python / central_server.py"]
        F -->|Real-Time Inference| G["Isolation Forest (Unsupervised ML)"]
        F -->|Data Historian| H[("Local CSV Logging (public/motor_parameters.csv)")]
    end
    
    subgraph Control Room [3D Spatial Frontend]
        H -->|High-Frequency Playback Feed| I["React 3D Command Dashboard"]
        I -->|React Three Fiber (R3F)| J["3D Digital Twin Spatial Factory Map"]
    end
    
    G -->|Triage & Emergency Alerts| I

Loading

---

⚡ Core Engineering Pillars

• Tri-Modal Sensor Fusion: Correlates power stability, current draws, and physical acoustics to prevent expensive, false-positive factory shutdowns caused by simple mechanical bumps.
• Unsupervised Anomaly Isolation: Employs an edge-optimized Isolation Forest algorithm that learns the unique, healthy signature of each motor on-the-fly, completely bypassing the need for manual training labels or historic failure logs.
• 3D React Three Fiber (R3F) Digital Twin: Renders a gorgeous, zero-latency blueprint factory grid. Anomaly states (>70% confidence) trigger real-time Red visual pulses and dynamic spin accelerations directly on the failing 3D motor node.
• Decoupled MQTT Scalability: Add or remove dozens of physical motor nodes instantly. The Mosquitto MQTT broker routes incoming telemetry seamlessly without requiring a single central codebase alteration.

---

🔬 Rule-Based Diagnostic Pipeline

  Telemetry Ingestion (Current + Voltage + Vibration)
       │
       ├──► [High Vibration]  + [Steady Current]  ==► External Mechanical Noise  (Filter & Ignore)
       ├──► [High Current]    + [Low Voltage]     ==► Grid Sag / Brownout Warning (Log System State)
       └──► [High Current]    + [High Vibration]  ==► MOTOR FAULT / TRIP TRIGGER  (Emergency Shutdown)

---

⚙️ Specifications & Local Constraints

• Low Computational Footprint: Engineered to run optimally on standard edge industrial computers (e.g., dual-core MacBook Air 2017 i5, 8GB RAM).
• Memory Footprint: Central Python server runs on <50MB RAM with zero external heavy ML framework dependencies (e.g., TensorFlow, PyTorch).
• Zero-GPU Rendering: React 3D Digital Twin uses highly optimized WebGL shaders and lightweight R3F meshes, maintaining a solid 60 FPS on standard integrated graphics.

---

🚀 Quick Start (Under 60 Seconds)

1. Install & Seed Python ML Backend

Ensure you have Python 3.8+ installed, then run the simulation and data generation script:

# Install Python analytical dependencies
pip install streamlit pandas numpy scipy scikit-learn pyserial plotly

# Generate mock physical telemetry data
python generate_motor_data.py

2. Launch 3D Digital Twin Frontend

Move to the project root, install packages, and spin up the Vite development server:

# Install Node packages
npm install

# Start Vite React Dashboard
npm run dev

3. Setup MQTT Broker (Optional Production Setup)

Start Mosquitto locally to route real MQTT data packets from physical ESP8266 controllers:

# Debian / Ubuntu
sudo apt install mosquitto mosquitto-clients -y
sudo systemctl start mosquitto

# Publish test JSON payload to verify connection
mosquitto_pub -h localhost -t "factory/motor1/telemetry" -m '{"voltage": 220, "current": 4.2, "vibration": 0.08}'

---

🖥️ Command Center & Visual Triage

• Blueprint Factory Floor: Minimalist off-black canvas styled specifically to minimize eye fatigue for factory control room operators.
• Triage Automation Panel: Failing motors are automatically isolated and pushed to the top of the operator queue for rapid physical inspection.
• Historian Exporting: Download entire filtered, raw sensor-fusion datasets instantly back to CSV format for post-mortem analysis.

---

📄 License

Distributed under the MIT License. See LICENSE for details.