PMSM的基于神经网络非线性预测控制  

Neural network based nonlinear predictive control for PMSM

作　　者：许泽伟[1] 胡玉娥[1] 李书臣[1] 

机构地区：[1]辽宁石油化工大学信息与控制工程学院,辽宁抚顺113001

出　　处：《贵州大学学报（自然科学版）》2008年第4期346-349,共4页Journal of Guizhou University:Natural Sciences

基　　金：辽宁省教育厅科学研究计划资助(2004D031)

摘　　要：采用三层前向神经网络作为PMSM转速预测模型,对PMSM进行非线性预测控制。先通过离线训练获得初始预测模型,再在线对神经网络模型的权值和阀值进行调整;控制算法是将预测模型处理成线性和非线性两部分,并用一种快速线性预测控制方法求取控制律,大大减小在线计算量和提高控制的实时性。最后仿真结果表明该方法具有良好的动、静态特性和抗干扰能力。Based on adopted three-layer feedforward neural network as the PMSM speed predictive model, a nonlinear predictive control for PMSM is presented in this paper. The primary predictive model is obtained using offline training method, and online training is carried out to update the weights and biases. The predictive model is separated into a linear part and nonlinear part by the control algo- rithm, and the control law is gotten by using a simple linear quick predictive control methods. Simulation results show that this method has good performance characteristic in transient and steady states, and also at variable-speed operation or load variation.

关 键 词：永磁同步电动机 神经网络模型 非线性预测控制 

分 类 号：TP273.5[自动化与计算机技术—检测技术与自动化装置]