PMSM伺服系统的神经网络动态面控制  

Neural Network Dynamic Surface Control of PMSM Servo System

作　　者：吴峰[1] 于洋[1] WU Feng;YU Yang(School of Electrical Engineering,Liaoning University of Technology,Jinzhou 121001)

机构地区：[1]辽宁工业大学电气工程学院,锦州121001

出　　处：《系统科学与数学》2021年第5期1203-1214,共12页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金(61603165);辽宁省自然科学基金(2019-BS-119)资助课题。

摘　　要：针对永磁同步电动机(PMSM)伺服系统中存在负载扰动和参数不确定问题,提出一种新型的基于预估器的自适应神经网络动态面控制算法.采用神经网络在线辨识系统的不确定项.在控制设计中引入神经网络预估器,并利用预估偏差学习神经网络的权值向量.通过引入动态面控制技术克服传统反步递推控制设计中存在的"复杂性爆炸"问题.所提控制算法可以在实现快速自适应的同时,避免了控制输入中可能存在的高频信号和控制输入漂移问题,进而提高PMSM伺服系统的位置跟踪控制性能.仿真结果验证了所提控制算法的有效性,并且与现有自适应神经网络控制方案相比,基于文章提出的自适应神经网络动态面控制方案的PMSM伺服系统具有更好的位置跟踪性能.In this paper,a novel adaptive neural network dynamic surface control(NNDSC)algorithm is proposed for the permanent magnet synchronous motor(PMSM)servo system considering parameter uncertainties and load torque disturbances.NNs are utilized to identify the uncertain terms.Predictors are introduced in the control design,and the predictor errors are used to learn the weight vectors of the NNs.DSC technique is introduced to overcome the"explosion of complexity"problem inherent the traditional backstepping design.The proposed control algorithm can achieve fast learning without high-frequency signals and zero drift in control input,which can improve the position tracking control performance of the servo system.Simulation results are provided to verify the effectiveness of the proposed control algorithm in this paper.Moreover,compared with existing adaptive NN control schemes,the PMSM servo system based on the proposed adaptive NNDSC achieves better position tracking control performance.

关 键 词：永磁同步电动机 自适应控制 神经网络控制 动态面控制 

分 类 号：TM341[电气工程—电机] TP183[自动化与计算机技术—控制理论与控制工程]