摘要：为了获取开关磁阻电机(SRM)的精确模型，将SRM的建模问题转化为一种非线性约束优化问题进行模型参数辨识研究，对传统的最小二乘支持向量机(LSSVM)回归算法进行改进，提出了惩罚因子自选取的LSSVM参数辨识算法，并给出了算法的具体实现步骤。设计磁链检测装置获取样机的磁化曲线，利用该实验数据并根据所提算法来辨识SRM模型参数。针对一台12/8极SRM实验样机开展了仿真与实验研究，仿真与实验结果的对比表明该辨识模型精度较高，能够较好地反映SRM的实际工作情况，同时验证了本文的研究方法合理有效。

关键词：开关磁阻；最小二乘支持向量机；参数辨识；磁链

Abstract: Parameter identification for optimization problems under nonlinear constraint conditions was used into modeling of switched reluctance motors(SRM), in order to obtain precise models of SRM. The conventional least suqares support vector machine(LSSVM) regression algorithm was improved, and then the LSSVM identification algorithm and its realization were proposed with the penalty parameter selfselected. SRMs model parameters were identified by the forenamed algorithm and the flux characteristic tested by the detection device. Taking a 12/8 poles SRM for example, the simulation and experiment were developed. The comparison between the simulated and the experimental results has verified this identified models accuracy and this identificaiton methods effectiveness.

Key words: switched reluctance; least squares support vector machine; parameter identification; fluxlinkage