Hybrid Intelligent Controller Design for Three-Disc AFPMSM in Electric Vehicles

Abstract

Abstract: This study introduces a torque distribution control system for a three-disc axial flux permanent magnet synchronous motor (AFPMSM) using a genetic algorithm-optimized back-propagation neural network (BP_ANN_GA) and an Adaptive Neuro-Fuzzy Inference System (ANFIS)-based torque control. These controllers enhance torque accuracy, reduce energy losses, improve motor stability and performance, and meet the demands of modern EVs by advancing efficient transmission technology. The BP_ANN_GA controller predicts torque distribution (Tm1, Tm2, Tm3) based on speed and required torque (T*m), with a genetic algorithm optimising the network’s weights and biases to minimise mean squared error (MSE) for greater accuracy. The ANFIS controller combines fuzzy logic for stator current control with real-time ANN adjustments to minimise stator current errors using a 5x5 fuzzy rule matrix, regulating the inverter output voltage. This integration adapts to dynamic driving conditions, improving vehicle performance and energy efficiency. MATLAB/SIMULINK simulations validate the system, showing balanced torque distribution, reduced energy losses, enhanced drivetrain efficiency, and strong adaptability to load or road changes, ensuring stability and superior dynamic response.