在线递推支持向量机的磁轴承转子位移预测

Rotor Displacement Prediction for Magnetic Bearings Based on On-Line Recursive Least Squares Support Vector Machine

研究了一种基于在线递推支持向量机的三相交流主动磁轴承转子位移实时预测方法。阐述了算法在线学习原理,建立了转子位移在线预测模型。该模型通过在线学习,形成理想的支持向量机结构,实现等效电流、等效磁链与转子位移的非线性映射,达到磁轴承转子位移的在线预测。仿真和实验结果表明所提算法具备较高的预测精度和较强的适应性。

A rotor displacement real- time prediction method for active magnetic bearings using on- line recursive least squares support vector machine（ LS- SVM） was proposed in this paper. The principle of on- line recursive LS-SVM was introduced,and the rotor displacement prediction model of a radial magnetic bearing was built. In this model,a better SVM structure was built through the on- line learning to form an efficient nonlinear mapping between suspension currents,magnetic flux linkage and rotor displacement,so that,it can realize the rotor displacement online prediction. To verify the effectiveness of the proposed method,some simulations and experiments were carried out. The findings show that the prediction model based on on- line recursive LS- SVM has a high precision and strong adaptability.