摘要
为实现三自由度混合磁轴承转子位移自检测,提出了基于粒子群优化最小二乘支持向量机的转子位移预测建模方法。通过对该磁轴承电磁结构和工作原理的分析,基于等效磁路法构建了大气隙范围内的非线性模型。在此模型基础上,结合最小二乘支持向量机在有限样本下对高维非线性的拟合及预测能力,通过采集具有代表性的电流–位移样本数据,训练得到磁轴承位移预测模型。针对最小二乘支持向量机超参数选取问题,采用粒子群优化算法进行自动寻优,以提高预测模型的拟合和预测精度。最后将均值误差和绝对误差作为模型评价指标对所提方法进行对比仿真研究,并对结果进行了讨论,验证了预测建模和自检测方法的有效性。
To realize the rotor displacement self-sensing for a 3-degree-of-freedom hybrid magnetic bearing（3-DOF-HMB）.This paper presents a novel predictive modeling method of rotor displacement for 3-DOF-HMB using particle swarm optimized-least squares support vector machines（PSO-LS-SVM）.First,the structure and working principle of 3-DOF-HMB is explained and the nonlinear mathematical model in the large air-gap is derived.Then,through the collection of representative current displacement data based on the nonlinear model,the predictive model of 3-DOF-HMB is obtained by training LS-SVM.Besides,the PSO algorithm is used to optimize parameters of LS-SVM to improve the performance of the predictive model.Finally,take mean squared error（MSE） and absolute error（AE） as model evaluation index to conduct comparative simulation research.The result is discussed and the effectiveness of predictive modeling and self-sensing method is verified.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2012年第33期118-123,184,共6页
Proceedings of the CSEE
基金
国家自然科学基金项目(61074019)
江苏省研究生创新计划(CXZZ12_0686)~~
关键词
混合磁轴承
支持向量机
粒子群优化
预测建模
hybrid magnetic bearing
support vector machine
particle swarm optimization
predictive modeling
