Digital Twin for Predictive Maintenance

Factories lose crores to unplanned machine breakdowns — a bearing fails without warning and the whole line stops. Industry 4.0's answer is the digital twin: a live virtual replica of the physical machine, fed by sensors, that predicts failures before they happen. This project builds exactly that on a demo motor: vibration and temperature sensors stream via ESP32 into a 3D dashboard twin, and an ML anomaly model flags bearing degradation days in advance. It's the only credible Industry 4.0 kit topic for mechanical and mechatronics students, it captures the "predictive maintenance using ML" search volume too, and no kit seller on the internet has a purchasable digital-twin page.

1. An accelerometer and temperature sensor mounted on the demo motor stream data to the ESP32.
2. The ESP32 publishes vibration windows and temperature readings to the cloud via MQTT.
3. The dashboard renders the 3D motor twin, updating rotation, heat tint, and vibration gauges in real time.
4. The ML model (autoencoder trained on healthy-run data) scores each vibration window; rising reconstruction error signals developing faults.
5. When the health score crosses warning/critical thresholds, alerts fire and the twin highlights the suspect bearing — days before a real failure.

• Live 3D dashboard twin of the demo motor (Three.js) mirroring real sensor state
• Vibration (accelerometer) + temperature sensing via ESP32 at 100+ Hz sampling
• ML anomaly detection: autoencoder trained on healthy vibration signatures flags degradation
• Remaining-useful-life estimate with trend charts as vibration drifts from baseline
• Threshold alerts (SMS/dashboard) when vibration or temperature crosses safe limits
• Historical playback: scrub back through the twin's recorded states
• Fault-injection demo: introduce imbalance to watch the twin and model react live
• Exportable health reports with FFT spectra for the project report

• ESP32 firmware + dashboard + ML pipeline source code
• Sensor mounting rig and demo motor assembly
• Pre-trained anomaly model + healthy-baseline dataset
• Project report PDF (synopsis, digital-twin literature, methodology, results, conclusion)
• Presentation PPT for review/demo day
• Viva Q&A sheet covering digital twins, vibration analysis, FFT, and autoencoders
• Demo script: fault injection and prediction walkthrough