An Adaptive Fault Diagnosis of Electric Vehicles: An Artificial Intelligence Blended Signal Processing Methodology

Abstract

This article demonstrates an innovative design of a sensorless technique to diagnose, monitor, and broadcast faults in an electric vehicle’s (EV) propulsion operating conditions. By utilizing the artificial intelligence with a signal processing mixed clustering technique, an onboard health monitoring system (HMS) has been presented. The clustering technique uses a data-mining approach to prevent future failures for predictive maintenance planning, which is novel. For example, the propulsion inverter is equipped with a diagnostic system that uses the proposed algorithm to compare the reference gate-driving signal with the actual output voltage of the voltage source inverter (VSI). This article presents different failure scenarios of the inverter and demonstrates the capability to be applied to other components, such as brakes and motors. To validate the proposed technique, the necessary algorithm calculations, simulation, and laboratory prototype results are provided. The proposed work is proven accurate with fast response in healthy and faulty conditions.