基于径向基神经网络的压电作动器建模与控制  

Modeling and control of piezoelectric actuator based on radial basis function neural network

作　　者：范家华[1] 马磊[1] 周攀[1] 刘佳彬[1] 周克敏[2] FAN Jia-hua;MA Lei;ZHOU Pan;LIU Jia-bin;ZHOU Ke-min(Institute of Systems Science and Technology, School of Electrical Engineering,Southwest Jiaotong University, Chengdu Sichuan 611756, China;School of Electrical Engineering and Computer Science, Louisiana State University, Baton Rouge 70803, USA)

机构地区：[1]西南交通大学电气工程学院系统科学与技术研究所,四川成都611756 [2]路易斯安那州立大学电气工程与计算机科学系,美国巴吞鲁日70803

出　　处：《控制理论与应用》2016年第7期856-862,共7页Control Theory & Applications

基　　金：国家自然科学基金重点项目(61433011)资助~~

摘　　要：针对压电作动器(piezoelectric actuator,PEA)的率相关迟滞非线性特性,构建了Hammerstein模型对压电作动器建模.采用径向基(radial basis function,RBF)神经网络模型表征迟滞非线性,利用自回归历遍模型(auto-regressive exogenous,ARX)表征频率的影响,并对模型参数进行了辨识.此模型可以在信号频率在1～300 Hz范围内时,较好地描述压电作动器的迟滞特性,建模相对误差为1.99%～4.08%.采用RBF神经网络前馈逆补偿控制,结合PI反馈的复合控制策略实现跟踪控制,控制误差小于2.98%,证明了控制策略的有效性.For the rate-dependent hysteresis nonlinearity of piezoelectric actuators, a Hammerstein model is established.Using a radial-basis-function (RBF) neural network to represent the hysteresis nonlinearity, an auto-regressive exogenous (ARX) model to represent the impact of frequency, and parameter identification is also accomplished. The proposed model describes the hysteresis characteristics of frequency ranged from 1 to 300 Hz of the signals, and the relative error is 1:99%~ 4:08%. A compound control strategy with RBF neural network feedforward inverse compensation and PI feedback is utilized for position tracking control, and the relative error less than 2:98%. Validity of the control strategy is proved by experimental results.

关 键 词：率相关 迟滞 RBF神经网络 压电作动器 HAMMERSTEIN模型 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP273[自动化与计算机技术—控制科学与工程]