Accurate torque modeling with PSO-based recursive robust LSSVR for a segmented-rotor switched reluctance motor

Abstract

In order to improve the reliability in torque calculation of SRM, an accurate nonlinear torque model regresses by recursive robust least squares support vector regression (RR-LSSVR) is proposed in this paper. The model is in terms of a segmented-rotor switched reluctance motor (SSRM). The characteristics of the SSRM is introduced to show its nonlinear characteristics both in magnetic and torque. Then, its mathematic model is established, and an accurate inductance measurement method and a torque calculation method are presented. After this, the principle of the RR-LSSVR and why it can adjust weights according to errors are described. The model used the RR-LSSVR algorithm shows an outstanding capability in accuracy and quickness compared with other algorithms. Finally, to further validate the accuracy of the proposed model in practical application, simulation and experiment are designed based on a 16/10 SSRM.

Keywords

• switched reluctance motor
• intelligent algorithm
• nonlinear modeling
• torque calculation
• lssvr