Improved Direct Instantaneous Torque Control Strategy of Switched Reluctance Motor based on Artificial Neural Network

Torque ripple is one of the most significant issues preventing the use of switched reluctance motors in electric vehicle applications. The Direct Instantaneous Torque Control (DITC) strategy is considered an efficient and more reliable solution to address these torque ripple issues. However, in conventional DITC, negative torque and significant torque ripple are still unavoidable. As a result, this paper presents an improved DITC approach based on adaptive switching angles of a switched reluctance motor (SRM), which can effectively mitigate torque ripple. Firstly, an artificial neural network (ANN) with a modified particle swarm optimization (MPSO) has been proposed to optimize the switch-off angle under various operating conditions in order to prevent the motor from generating negative torque, which leads to minimizing the average torque and reducing efficiency. The MPSO algorithm is employed for training the ANN weights according to the multiobjective function to identify the optimum switch-off angle. Additionally, the switch-on angle is analytically adopted in each electrical cycle, depending on the flux derivation, to fully utilize the torque-producing capability. Finally, in order to prove the effectiveness of the proposed DITC approach, various simulation results are performed and compared with the traditional DITC method using a four-phase, 4KW, 8/6 SRM prototype.