Induction Motor Current Control with Torque Ripples Optimization Combining a Neural Predictive Current and Particle Swarm Optimization

This paper develops a predictive current controller for an induction motor (IM), based on neural networks. More precisely, the proposed regulator treats this control problem as an optimization problem exploiting a neural predictor for IM currents. The considered objective function is constituted of two components, namely: the current tracking errors and the electromagnetic torque ripples. Moreover, this objective function is computed over a given time horizon, based on the IM currents prediction results. The Particle Swarm Optimization (PSO) algorithm is used to solve this optimization problem. The paper provides simulation results using Matlab/Simulink to prove the effectiveness of the proposed regulator compared to a conventional controller (PI). Indeed, the simulation results show that the proposed controller offers several advantages, including robustness for load variations, rotor resistance variations, and DC bus voltage variations. In addition, these results highlight a reduction in torque ripples and overshoots during transient regimes.