基于小波神经网络和非线性扰动观测器的直线伺服系统控制  

Control of Linear Servo System Based on Wavelet Neural Network and Nonlinear Disturbance Observer

作　　者：赵希梅[1] 原浩 朱文彬 Zhao Ximei;Yuan Hao;Zhu Wenbin(School of Electrical Engineering Shenyang University of Technology ,Shenyang 110870 China)

机构地区：[1]沈阳工业大学电气工程学院

出　　处：《电工技术学报》2019年第19期3989-3996,共8页Transactions of China Electrotechnical Society

基　　金：辽宁省自然科学基金计划重点项目（20170540677）;辽宁省教育厅科学技术研究项目（LQGD2017025）资助

摘　　要：针对永磁直线同步电机(PMLSM)易受外部负载扰动、参数变化和摩擦力等非线性不确定性因素影响而导致伺服系统性能降低的问题,提出一种基于小波神经网络(WNN)的非线性扰动观测器(NDO)控制方法。首先,将非线性模型线性化,然后利用线性系统理论设计反馈线性化控制器(FLC),实现位置跟踪,从而使 PMLSM 控制系统稳定;采用 NDO 估计并补偿系统的不确定性,降低了系统跟踪误差。但是在实际运行过程中观测器增益较难选取,极易产生较大的观测误差,为了增强系统鲁棒性,通过 WNN 在线补偿 NDO 的观测误差,以改善 NDO 的补偿能力。通过系统实验,证明所提出方法的有效性,系统具有较强的鲁棒性和良好的跟踪精度,可以有效补偿系统存在的不确定性对系统跟踪性能的影响。A nonlinear disturbance observer (NDO) control method based on wavelet neural network (WNN) is used to improve the performance of the permanent magnet linear synchronous motor (PMLSM), which is easily influenced by nonlinear uncertainties such as external load disturbance, parameter change and friction, etc. Firstly, linearize the nonlinear model. Then, the linear system theory is used to design the feedback linearization controller (FLC) for the main position tracking to make the PMLSM control system stable. The uncertainties of the system is estimated and compensated by NDO, and the system tracking error is reduced. However, in the actual operation, the observer gain is difficult to select, and it is very easy to produce large observation errors. In order to enhance the robustness of the system, compensating observation errors online through WNN to improve the compensation ability of NDO. The experimental results show that the proposed method is effective, the system has strong robustness and good tracking accuracy, which can effectively compensate for the uncertainties existing in the system.

关 键 词：永磁直线同步电机 反馈线性化控制器 非线性扰动观测器 小波神经网络 

分 类 号：TP351[自动化与计算机技术—计算机系统结构] TM359.4[自动化与计算机技术—计算机科学与技术]