🔌 DCRM Fault Detection System

AI-Powered Circuit Breaker Monitoring & Analysis Platform

---

📋 Table of Contents

• Problem Statement
• Solution Overview
• Key Features
• Technology Stack
• System Architecture
• Installation Guide
• Usage
• API Documentation
• Screenshots
• Future Scope
• Team
• License

---

🎯 Problem Statement

Circuit breaker faults in electrical distribution systems lead to:

• ❌ Unplanned power outages
• ❌ Equipment damage and safety hazards
• ❌ High maintenance costs
• ❌ Lack of predictive maintenance capabilities

Traditional DCRM (Direct Contact Resistance Measurement) testing is manual, time-consuming, and reactive rather than proactive.

---

💡 Solution Overview

Our AI-Powered DCRM Fault Detection System revolutionizes circuit breaker monitoring by:

✅ Automated Analysis - Instant fault detection using machine learning
✅ Real-Time Monitoring - 24/7 continuous surveillance with live dashboards
✅ Predictive Maintenance - Early fault detection before failures occur
✅ Multiple Interfaces - Web-based analysis + Live monitoring dashboard
✅ High Accuracy - 100% accuracy on test datasets with confidence scoring

---

✨ Key Features

🔍 Analysis Module

• CSV File Upload - Drag & drop support for historical data analysis
• Manual Data Input - Paste comma-separated resistance values
• Test Sample Generation - Generate synthetic waveforms for testing
• Instant Results - Fault detection with confidence scores and extracted features

📡 Real-Time Monitoring Dashboard

• Live Waveform Visualization - Chart.js powered real-time graphs
• Animated Circuit Breaker - Visual representation with state indicators
• Fault Detection Display - Color-coded alerts with confidence metrics
• Statistics Panel - Total scans, faults detected, uptime, health rate
• Alert History - Comprehensive log of all detected anomalies
• Adjustable Monitoring Speed - Configurable update intervals (0.5s - 5s)
• Fault Simulation - Test with Normal, Spike, Plateau, Unstable signals

🤖 AI/ML Capabilities

• Random Forest Classifier - 200 estimators with optimized parameters
• Feature Extraction - 9 key features including peaks, slopes, plateaus
• Multi-Class Classification - Detects Normal, Spike, Plateau, Unstable faults
• Confidence Scoring - Probability distribution for all fault types
• Trained on Synthetic Data - 300 samples with noise and variations

🎨 User Interface

• Modern Glassmorphism Design - Frosted glass effects with backdrop blur
• Animated Backgrounds - Floating gradient orbs and moving grid patterns
• Color-Coded Themes - Purple/Pink for analysis, Teal/Cyan for monitoring
• Responsive Design - Mobile, tablet, and desktop optimized
• Smooth Animations - Hover effects, transitions, loading states

---

🛠️ Technology Stack

Backend

• Python 3.8+ - Core programming language
• Flask 2.3+ - Web framework for REST API
• Flask-CORS - Cross-origin resource sharing
• Scikit-learn - Machine learning library
• NumPy - Numerical computations
• SciPy - Signal processing
• Joblib - Model persistence

Frontend

• HTML5 - Semantic markup
• CSS3 - Modern styling with animations
• JavaScript (ES6+) - Dynamic functionality
• Chart.js - Real-time waveform visualization

AI/ML

• Random Forest Classifier - Ensemble learning method
• Feature Engineering - Peak detection, statistical analysis
• Signal Processing - Savitzky-Golay filtering

Future Integration

• ESP32 Microcontroller - Hardware data acquisition
• IoT Sensors - Real-time DCRM measurements
• Cloud Deployment - Heroku, Vercel, Railway

---

🏗️ System Architecture

┌─────────────────────────────────────────────────────────┐
│                    User Interface Layer                  │
├──────────────────────┬──────────────────────────────────┤
│   Analysis Page      │   Monitoring Dashboard           │
│   (index_v2.html)    │   (monitor_v2.html)             │
│   - Upload CSV       │   - Live Waveform Chart         │
│   - Manual Input     │   - Circuit Breaker Visual      │
│   - Test Generation  │   - Real-time Statistics        │
└──────────┬───────────┴──────────────┬──────────────────┘
           │                          │
           ▼                          ▼
┌─────────────────────────────────────────────────────────┐
│                    Flask REST API                        │
│                 (dcrm_flask_api.py)                     │
├─────────────────────────────────────────────────────────┤
│  Endpoints:                                             │
│  - POST /predict   → Analyze waveform                   │
│  - GET  /health    → API health check                   │
│  - GET  /          → API information                    │
└──────────┬──────────────────────────────────────────────┘
           │
           ▼
┌─────────────────────────────────────────────────────────┐
│              AI Processing Layer                         │
├─────────────────────────────────────────────────────────┤
│  1. Feature Extraction                                  │
│     - Peak detection (SciPy)                            │
│     - Statistical metrics (NumPy)                       │
│     - Plateau duration analysis                         │
│     - Slope calculations                                │
│                                                          │
│  2. ML Model (Random Forest)                            │
│     - 200 decision trees                                │
│     - 9 input features                                  │
│     - 4 output classes                                  │
│     - Confidence scoring                                │
└──────────┬──────────────────────────────────────────────┘
           │
           ▼
┌─────────────────────────────────────────────────────────┐
│                 Data Sources                             │
├─────────────────────────────────────────────────────────┤
│  Current: Synthetic Data Generation                     │
│  Future:  ESP32 → DCRM Sensors → WiFi → API            │
└─────────────────────────────────────────────────────────┘

---

📦 Installation Guide

Prerequisites

• Python 3.8 or higher
• pip package manager
• Modern web browser (Chrome, Firefox, Edge)

Step 1: Clone Repository

git clone https://github.com/yourusername/dcrm-fault-detection.git
cd dcrm-fault-detection

Step 2: Install Python Dependencies

cd backend
pip install flask flask-cors numpy scipy scikit-learn joblib

Or use requirements.txt:

pip install -r requirements.txt

Step 3: Train AI Model (Optional)

python train_dcrm_classifier.py

This generates dcrm_fault_classifier_v2.joblib

Step 4: Start Flask API

python dcrm_flask_api.py

API starts at http://localhost:5000

Step 5: Open Frontend

Navigate to frontend/ folder and open index_v2.html in your browser.

Or use a local server:

cd frontend
python -m http.server 8000

Then visit http://localhost:8000

---

🚀 Usage

Analysis Mode

1. Upload CSV File

   • Click browse or drag & drop your CSV file
   • File should contain resistance measurements
   • Click "Analyze CSV" button

2. Manual Input

   • Paste comma-separated resistance values
   • Minimum 10 values required
   • Click "Analyze Data" button

3. Generate Test Sample

   • Select fault type (Normal, Spike, Plateau, Unstable)
   • Click "Generate & Analyze"
   • System creates and analyzes synthetic waveform

Real-Time Monitoring Mode

1. Click "📡 Open Live Monitoring Dashboard" button
2. Adjust monitoring speed (0.5s - 5s intervals)
3. Select fault simulation type
4. Click "🚀 Start Monitoring"
5. Watch live analysis with:
   • Real-time waveform chart
   • Animated circuit breaker
   • Fault detection alerts
   • Statistics updates
   • Alert history

---

📡 API Documentation

Base URL

http://localhost:5000

Endpoints

1. Get API Information

GET /

Response:

{
  "message": "DCRM Fault Detection API",
  "version": "1.0",
  "endpoints": {
    "/predict": "POST - Analyze waveform",
    "/health": "GET - Check API health"
  }
}

2. Health Check

GET /health

Response:

{
  "status": "healthy",
  "model_loaded": true
}

3. Predict Fault

POST /predict
Content-Type: application/json

Request Body:

{
  "waveform": [2.5, 2.6, 2.4, 2.7, ...]
}

Response:

{
  "success": true,
  "predicted_fault": "spike",
  "confidence": {
    "normal": 0.00,
    "spike": 77.00,
    "plateau": 8.50,
    "unstable": 14.50
  },
  "features_extracted": {
    "num_peaks": 1,
    "max_peak_height": 1.000,
    "mean": 0.089,
    "std": 0.110,
    "plateau_duration": 10,
    "max_slope": 0.3872,
    "min_slope": -0.2341
  }
}

---

📸 Screenshots

Analysis Page

• Modern UI with gradient animations
• Three upload options
• Instant results display

Real-Time Monitoring Dashboard

• Live waveform chart
• Animated circuit breaker
• Statistics panel
• Alert history

Fault Detection Results

• Confidence scores with progress bars
• Extracted features display
• Color-coded fault indicators

---

🔮 Future Scope

Short-term Enhancements

• ✅ ESP32 hardware integration
• ✅ WebSocket support for real-time data streaming
• ✅ Database integration (PostgreSQL/MongoDB)
• ✅ User authentication and multi-user support
• ✅ Historical data analytics and trends

Long-term Vision

• 🚀 Mobile app (Flutter/React Native)
• 🚀 Advanced ML models (LSTM, Transformer)
• 🚀 Predictive maintenance scheduling
• 🚀 Multi-sensor support (temperature, vibration)
• 🚀 Cloud deployment with auto-scaling
• 🚀 Integration with SCADA systems
• 🚀 Anomaly detection for multiple fault types
• 🚀 Report generation and export features

---

---

📄 License

This project is developed for Smart India Hackathon 2025.

---

🙏 Acknowledgments

• Smart India Hackathon organizing committee
• RGIPT, Jais
• Open-source libraries: Flask, Scikit-learn, Chart.js
• Inspiration from modern UI design trends

---

📞 Contact

For queries related to this project:

• Email: akhawattushar@gmail.com
• GitHub: https://github.com/akhawattushar
• LinkedIn: www.linkedin.com/in/tushar-singh-akhawat-9a4848341

---

Made with ❤️ for Smart India Hackathon 2025