Study on collaboration of a machine learning-based fault detector and a fault tolerant direct yaw moment controller for distributed drive electric vehicles

Distributed drive electric vehicles (DDEVs) equip independent motors on all drive wheels. Therefore, faulty motors can significantly impact the operational efficiency and safety of these vehicles. The impact is especially notable under challenging driving conditions, such as slippery roads, compared to normal driving conditions. Hence, to ensure comfortable and safe driving conditions, it is crucial to explore fault detection and fault-tolerant control systems. This study introduces a novel collaborative system that integrates an online fault detector, based on support vector machines (SVMs), with a fault-tolerant controller utilizing direct yaw moment control. The proposed system efficiently identifies types and locations of motor faults and dynamically adjusts the torque output of the functioning motors to mitigate various faulty motor scenarios. The fault detector is trained offline using SVM using phase current signals of the motors. SVM is chosen because it requires less training data compared to other machine learning techniques. The proposed fault-tolerant controller utilizes sliding mode control (SMC) for effective direct yaw moment control of the DDEV. The effectiveness of the proposed system is validated through simulations and prototype testing.