论文部分内容阅读
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。
In order to realize self-testing rotor displacement of three degrees of freedom hybrid magnetic bearings, a rotor displacement prediction modeling method based on Particle Swarm Optimization (LS-SVM) is proposed. By analyzing the electromagnetic structure and working principle of the magnetic bearing, a nonlinear model in the air gap was constructed based on the equivalent magnetic circuit method. Based on this model, combining with the fitting and prediction ability of least squares support vector machine (SVM) for high dimensional nonlinearity under limited sample, a magnetic bearing displacement prediction model is trained by collecting representative current-displacement sample data. In order to solve the problem of hyperparameter selection in least square support vector machines, particle swarm optimization (PSO) algorithm is used to optimize the parameters automatically to improve the fitting and prediction accuracy of the prediction model. Finally, the mean error and the absolute error are taken as the evaluation indexes of the model to simulate the proposed method. The results are discussed and the effectiveness of the predictive modeling and the self-testing method is verified.