基于主元分析神经网络补偿的感应电动机逆解耦控制

Inverse Decoupling Control of Induction Motor Based on PCA-NN Compensation

对于具有多变量、非线性、强耦合、慢时变等特征的异步电动机调速系统,实现定子磁链与电磁转矩的高精度动态解耦是提高系统性能的关键。首先通过非线性状态反馈建立感应电动机的α阶积分逆模型,并分析非线性状态反馈的误差对其逆模型精度的影响。在此基础上提出了一个基于主元分析神经网络补偿的感应电动机逆解耦控制方法,将补偿后的α阶积分逆模型串联到对象的输入端建立广义被控对象。复杂的感应电动机调速系统被解耦成电磁转矩与定子磁链的两个独立回路,利用线性系统理论分别对独立回路进行综合设计,实现定子磁链和电磁转矩对各自给定值的渐进跟踪。实验结果验证了建议方案的有效性和可行性。

It is important to realize torque and stator flux dynamic decoupling control for induction motor（IM） drive,which is multi-input and multi-output,nonlinear and strong coupling system.First,an ？-order integral inverse model of IM is constructed by the nonlinear state feedback,and the reason that the inverse model accuracy is affected by the feedback error is deeply analyzed.On the base of this,an inverse decoupling control method for IM based on principal component analysis neural network（PCA-NN） compensation is developed,in which the compensated ？-order integral inverse model is cascaded with the plant.The complex IM drive system is decoupled into two separate loops,i.e.torque loop and flux loop.The separate loop is respectively designed by linear system theory so that the asymptotic tracking of stator flux and torque is realized.The experimental results verify the validity of the proposed scheme.