基于最小二乘支持向量机的开关磁阻电动机建模.pdfVIP

27 卷 6 期 中 国 电 机 工 程 学 报 Vol.27 No.6 Feb. 2007 2007 年2 月 Proceedings of the CSEE 2007 Chin.Soc.for Elec.Eng. 文章编号：0258-8013 (2007) 06-0026-05 中图分类号：TM352 文献标识码：A 学科分类号：47040 基于最小二乘支持向量机的开关磁阻电动机建模 司利云，林 辉，刘 震 (西北工业大学自动化学院，陕西省 西安市 710072) Modeling of Switched Reluctance Motors Based on LS-SVM SI Lin-yun, LIN Hui, LIU Zhen (Automatic College Northwestern Polytechnical University, Xi ’an 710072, Shaanxi Province, China) ABSTRACT: Considering nonlinear magnetization 关键词：开关磁阻电动机；非线性磁特性；最小二乘支持向 characteristics of a switched reluctance motor (SRM), the paper 量机；建模；支持向量回归 first presents least squares support vector machine(LS-SVM) as 0 引言 a new tool to develop the model of the SRM. The basic premise of the LS-SVM regression is that it forms a very efficient 开关磁阻电机(SRM)虽然结构简单，但由于其 mapping structure for the nonlinear SRM. The sampled data set 本身的非线性电磁特性以及多变量强耦合的特点， obtained from the experimental SRM by the finite elements 使得用传统电机的性能分析很难得到有效的建模方 method(FEM), is comprised of magnetization data with 法。虽然可以从理论上列出1 个精确的数学表达式， position and current as inputs, and the corresponding 完整，准确地描述开关磁阻电机的电磁及 学的非 flux-linkage as output. Given a sufficiently large training data 线性关系，但计算却相当繁琐，特别是作为SRM 各 set, the LS-SVM can build up a correlation among position, 种性能计算方法基础的磁链–电流–位置的磁化曲线 current and flux-linkage. Compared with the models based on 族，由于相电流i 和转子位置角q之间的非线性耦合， artificial neural networks(ANN) methods, the proposed model 难以获得电感L 和磁链y 的解析解，从而使得这样精 has better capability of generalization and better convergent 确的模型无法直接进行实际应用。 speed if a small data set