摘要
针对开关磁阻电机电磁特性的非线性造成其精确数学模型难以建立的问题,根据样机实测自感特性和矩角特性,采用最小二乘支持向量机(LS-SVM)的方法对开关磁阻电机非线性建模。与几种常用的神经网络建模方法相比,采用LS-SVM方法所建模型学习速度快,模型精确度高。以一台8/6极开关磁阻电机为例,把基于LS-SVM的自感模型和矩角模型应用于开关磁阻电动机调速系统的建模并进行仿真。实验结果表明,相同工作条件下的仿真电流波形与实验电流波形基本一致,误差小于5%,说明该建模方法的正确性,为开关磁阻电动机调速系统的智能控制提供理论和实践参考。
The precise mathematical model of switched reluctance motor（SRM） is difficult to be estab-lished because of its nonlinear reluctance characteristic. In view of this problem,the switched reluctance motor was modeled by least squares support vector machine （LS-SVM） method according to the measured self-inductance and torque-angle characteristics. Compared with the model trained by several common neural network methods,the model based on LS-SVM is of superior learning speed and higher model ac-curacy. Taking an 8 /6 SRM for example,the self-inductance and torque-angle models based on LS-SVM were used to build the pattern of switched reluctance drive system and the simulation has been implemen-ted with Matlab. It has been shown by experiments that the current waveform obtained by simulation is consistent to the experimental current waveform and the error is less than 5% under the same operation condition. As a result,it has been proved that the LS-SVM modeling method is correct and effective, which provides the theoretical and practical reference for the intelligent control of switched reluctance drive system.
出处
《电机与控制学报》
EI
CSCD
北大核心
2010年第5期32-36,共5页
Electric Machines and Control
基金
国家"十一五"科技支撑计划重点项目(2007BAK29B05
2007BAB13B01)
关键词
开关磁阻电动机
神经网络
最小二乘支持向量机
非线性模型
矩角特性
switched reluctance motors
neural network
least squares support vector machine
nonlinear mathematical model
torque-angle characteristics
