Fault Detection Technique Based On Clustering Approach of Artificial Intelligence in Electric Vehicle Converters

An on-board health monitoring system to diagnose the vehicle while it is in operation; is a necessity to propose electric vehicles (EVs) as a robust replacement of the internal combustion engine (ICE) ones. This paper proposes an efficient, simple and reliable fault detection technique of an open circuit fault in voltage source inverters (VSI), as a case study. It is based on semi-clustering approach, which compares a gate-driving signal (as a reference) to the actual VSI output voltage, trained in many different scenarios. Simulink simulation model and a laboratory prototype are built to prove the validity of the proposed technique. Results are presented and discussed to confirm coherence of proposed methodologies.