Online Fault Detection in Traction PMSM Using a MEMS Accelerometer: A Deep Learning Approach

The modern Automotive sector widely adopts Permanent Magnet Synchronous Motors (PMSMs) for the propulsion of Electric Vehicles (EVs). As the market for EVs grows, reducing costs and increasing performance in terms of reliability, range, and efficiency becomes paramount. This work proposes a cost-effective, integrated fault detection system using a MEMS vibration sensor with 3.3 kHz bandwidth and an Autoencoder Neural Network. The model is deployed on the Microcontroller Unit (MCU) that drives the Voltage Source Inverter (VSI), a STM StellarE1 Automotive, utilizing 3 kiB of flash memory (0.97%) and 18 kiB of available RAM (0.6%). This system achieves an accuracy of the 98% in detecting the incipience of a natural bearing fault.