贝叶斯证据框架下LS-SVM的BPMSM磁链建模

Modeling of flux linkage for the BPMSM based on LS-SVM within the Bayesian evidence framework

针对常规解析法建立无轴承永磁同步电机(BPMSM)磁链模型的局限性,提出一种贝叶斯证据框架下最小二乘支持向量机(LS-SVM)的BPMSM磁链建模方法.对BPMSM磁链的非线性建模进行简单分析,在介绍LS-SVM回归理论和贝叶斯证据框架基本思想的基础上,通过贝叶斯证据框架推断准则1确定模型的权向量w,通过贝叶斯证据框架推断准则2确定模型的正则化参数γ,通过贝叶斯证据框架推断准则3确定模型的核参数σ,进而建立基于贝叶斯证据框架下LS-SVM的BPMSM磁链模型.在Matlab7.0环境下进行仿真研究.仿真结果表明,贝叶斯证据框架下LS-SVM的磁链模型具有拟合精度高、泛化能力强、结构灵活、计算速度快等特点.

According to the modeling of the flux linkage model for bearingless permanent magnet synchronous motors （BPMSMs） by using the conventional analytic method, a novel modeling method based on least squares support vector machine （LS-SVM） within the Bayesian evidence framework was proposed. Fistly, the nonlinear modeling of the flux linkage model was analyzed briefly. Secondly, the regression theory of the LS-SVM and the basic concept of the Bayesian evidence framework （namely, the three levels inference） were introduced in detail. And then the level 1 inference was used to determine the weight vector w of the LS-SVM. The level 2 inference was used to ascertain the model regularization parameter γ of the LS-SVM. The level 3 inference was used to obtain the kernel parameter a of the LS-SVM, and thus the flux linkage model of the BPMSM based on LS-SVM within the Bayesian evidence framework was established. Finally, the simulation studies were carried out with the Matlab7.0 software to illustrate the performance of the proposed method. Simulation results show that this model has high accurate precision, good generalization ability, flexible structure, and rapid calculation speed.