ANN-Based Fault-Tolerant Control of Open-End Winding Induction Motors for Electric Vehicles

This paper proposes a fault-tolerant operation (FTO) strategy for an open-end winding induction motor (OEWIM) drive powered by dual two-level voltage source inverters (VSIs). The proposed approach enhances the duty-ratiobased direct torque control (Duty-DTC) scheme by integrating an artificial neural network (ANN) and utilizing inherent redundancies among medium and small voltage vectors to maintain stable motor operation under single-inverter fault conditions. During healthy operation, the drive employs 19 distinct voltage vectors, which reduce to 7 in the presence of an inverter fault. A novel duty-ratio compensation method is introduced to dynamically regulate the magnitude of the applied voltage vector, thereby preserving the volt-second balance. Additionally, necessary modifications in the switching sequences and gating logic are detailed, alongside the adaptations in the offline-trained ANN architecture required for FTO. The proposed scheme is validated through MATLAB simulations and laboratory experiments.