Model Predictive Control of Switched Reluctance Machines With Online Torque Sharing Function Based on Optimal Flux-Linkage Curve

Abstract

Aiming at reducing the large ripple of torque of switched reluctance machines (SRMs), this paper introduces a model predictive control (MPC) approach that leverages the optimal flux-linkage curve for enhanced performance. According to the multi-objective optimization requirement, this paper regulates phase flux-linkage curve based on the Gravitational Search Algorithm (GSA). On this basis, the torque sharing function (TSF) is built according to optimal flux-linkage curve, thus improving the smoothness of flux-linkage curve and reducing ripple of torque. Finally, the TSF is combined with MPC to control torque. This method predicts the phase torque and phase current to construct a cost function and select the optimal candidate switch state to be applied to control the power converter, thus reducing the torque ripple. Simulations and experiments are performed on a three-phase 12/8-pole SRM to validate that the method proposed in this paper could effectually suppress the ripple of torque and reduce RMS (Root Mean Square) phase current.