🎯 What This Demo Shows

This is a Predictive Maintenance System for IoT appliances. It uses machine learning to detect degradation and predict when maintenance is needed.

☕ Smart Kettle Demo

• Simulate Boil: Each cycle adds mineral scale buildup
• Watch the health bar go from green → yellow → red
• Descale: Resets the kettle to healthy state
• Try 3× (30 cycles) to see severe scaling quickly

🍳 Kitchen Chimney Demo

• Simulate Usage: Cooking adds grease buildup
• AI Auto-Clean: When ON, automatically cleans at <30% health
• Toggle OFF to see full degradation manually
• Try 3× (30 cycles) repeatedly to see the curve

📊 ML Metrics Explained

• Health Score 100% = Perfect, 0% = Needs maintenance
• Slope Rate of degradation per cycle
• Acceleration Is degradation speeding up?
• RUL Predicted cycles until failure

🧠 How the ML Model Works

• Polynomial Regression (degree 2): Fits curve to last 20 data points
• Z-Score Anomaly Detection: Detects sudden spikes (threshold: 2.5σ)
• Health Score: Linear interpolation between baseline and failure threshold
• RUL: Extrapolates polynomial curve to predict future failure

💡 Demo Tips

• Start fresh: Descale/Auto-clean before demonstrating
• Use 3× speed for quick visible results
• Watch the chart update in real-time
• Compare with/without AI Auto-Clean enabled