Synchronous reluctance motor flux linkage saturation modeling based on stationary identification and neural networks

Abstract

For synchronous reluctance motor (SynRM) parameters are easy to change when the magnetic circuit is saturated, and the characteristics of significant nonlinearity, the control based on fixed parameters will easily lead to poor control accuracy. In this paper, the influence of magnetic saturation of the motor magnetic circuit on the parameters is taken into account, and the motor parameters based on the stationary identification technique are identified offline. Simultaneously, due to the effect of hysteresis saturation, it is difficult to simply fit the stationary identified magnetic flux linkage parameters into the motor parameter curves. This paper combines a neural network training model with a control algorithm to dynamically assign the optimal current in real-time to improve response speed and robustness when controlling synchronous reluctance motors. By applying the neural network training data to the simulation analysis of MTPA control, the effectiveness of the adopted identification method and motor modeling method as well as the identified parameters applied to the high-precision control of synchronous reluctance motors is verified.