Adaptive Neuro-Fuzzy Control of a Single-Sided AFPMSM Motor for Electric Vehicle Applications

This study will present the torque control design of an in-wheel single-sided AFPMSM motor utilizing Adaptive Neuro-Fuzzy (ANFIS) techniques powered by a two-level inverter. The neuro-fuzzy intelligent controller will be constructed using a neural network-based feature set and a fuzzy system. The neural network will process the dataset concerning the stator current error (e) and error integral (∆e). This dataset will be fed into the fuzzy logic approach to establish control rules. The ANFIS controller will undergo training and testing stages. An evaluation will compare the efficiency of the FLC, ANFIS torque, and PI controllers. The results of MATLAB/SIMULINK simulations illustrate that the proposed ANFIS torque controllers deliver precise control. The MATLAB/SIMULINK simulation results demonstrate that the ANFIS torque controllers exhibit greater precision than the PI controller. This study underscores the efficacy of employing the adaptive neuro-fuzzy technique, particularly with a two-level inverter, to ensure precise torque control for in-wheel single-sided AFPMSM motors. The amalgamation of neurofuzzy intelligent control, neural network-based feature sets, and fuzzy systems demonstrate promise in enhancing motor control design.