基于神经网络优化的交流自抗扰伺服系统控制  

Control of AC ADRC Servo System Optimized Based on Neural Network

作　　者：金爱娟[1] 陈昌泽 李少龙[1] JIN Ai-juan;CHEN Chang-ze;LI Shao-long(University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学,上海200093

出　　处：《包装工程》2021年第19期220-231,共12页Packaging Engineering

基　　金：国家自然科学基金(11502145)。

摘　　要：目的为了解决传统交流永磁同步电机伺服自抗扰控制系统中外界扰动、非线性特性和本身自抗扰控制中参数较多且整定难的问题。方法利用小波神经网络对自抗扰控制中的扩张状态观测器的误差校正系数进行在线整定,设计出基于小波神经网络优化的自抗扰控制器及相关的控制系统,以实现对整体控制系统的性能优化,并通过在Matlab/SIMULINK仿真实验与传统PID伺服控制系统和未进行优化的交流自抗扰伺服系统进行对比验证。结果仿真结果表明,基于小波神经网络优化的交流永磁同步电机伺服自抗扰控制系统对目标位置动态响应快、稳态误差小、抗干扰能力强,稳态时转矩脉动小。结论与常规未优化自抗扰伺服系统和传统PID伺服系统相比,基于小波神经网络优化后的自抗扰伺服系统,能有效地提高伺服系统控制性能和鲁棒性。The work aims to solve the problems of external disturbance and nonlinear characteristics in the active disturbance rejection control(ADRC) system of traditional AC permanent magnet synchronous motor servo and numerous parameters difficult to tune in ADRC. Wavelet neural network was used to adjust the error correction coefficients of the extended state observer in the active disturbance rejection control, so as to design the active disturbance rejection controller and related control system optimized based on wavelet neural network, thus optimizing the performance of the overall control system. Through Matlab/SIMULINK simulation experiment, the optimized system was compared with traditional PID servo control system and un-optimized AC ADRC servo system for verification. According to the simulation results, the ADRC system of AC permanent magnet synchronous motor servo optimized based on wavelet neural network had fast dynamic response to the target position, small steady-state error, strong anti-interference ability, and small steady-state torque ripple. Compared with conventional un-optimized ADRC system and traditional PID servo system, the ADRC system optimized based on wavelet neural network can effectively improve the control performance and robustness of servo system.

关 键 词：永磁同步电机 伺服系统 自抗扰控制 参数整定 小波神经网络 

分 类 号：TB486[一般工业技术—包装工程]