Embedded AI · 2026 · Lead Developer — Graduation Project (PFE)
Edge AI Predictive Maintenance System via ESP32 Multi-Sensor Array
ESP32CEdge AISignal ProcessingPredictive Maintenance
§ I
Abstract
A fully embedded predictive-maintenance pipeline running on an ESP32-S3, fusing MEMS vibration, current, and temperature sensing to detect early bearing and winding faults in three-phase induction motors.
§ II
System Architecture
ESP32-S3 → I²S MEMS accelerometer + Hall current sensor + NTC; FreeRTOS tasks for acquisition, FFT, and inference; 8-bit quantized autoencoder deployed via TFLite Micro.
§ III
Results
On-device inference in <12 ms per window, 96.4% F1 on synthetic fault dataset, total power draw under 320 mW.