Self-adaptive differential evolution algorithm with hybrid mutation operator for parameters identification of PMSM

Abstract

Parameters identification of permanent magnetic synchronous motor (PMSM), which significantly influences the control performance of the drive system, is an important and challenging task of power electronic system. The problem requires both high solution quality and fast convergence speed due to the constraints of hardware. This paper presents a self-adaptive differential evolution algorithm with hybrid mutation operator (SHDE) for parameters identification problem. In this method, a novel mutation operator, called “current-to-archive-best,” is developed by mixing the best solutions randomly selected from archive set and current population. Thus, the algorithm could use the best searching memories so far to generate promising solutions, yielding a faster evolving procedure. Besides, the corresponding control parameters of SHDE are also self-adapted without tedious trial-and-error progress to get appropriate values. Moreover, the parameters estimation program is inserted into the PMSM simulation that is solved by using Newton–Raphson method without any pre-assumption and simplification. This framework, which may be used under any working conditions with large disturbance, is different from other publications, resulting in wider applications. The proposed method applied to parameters identification of PMSM is evaluated on a PMSM drive system with two different operations. The comprehensive results and statistical analyses, compared with other state-of-the-art algorithms, show that SHDE could find high-quality solutions with higher convergence speed and probability.