含过程噪声的Hammerstein-Wiener神经模糊模型多信号源辨识  

Multi-Signal Identification of Hammerstein-Wiener Models Based on Bias Compensation Recursive Least Squares

作　　者：贾立[1] 杨爱华[1] 邱铭森[2] 

机构地区：[1]上海大学机电工程与自动化学院,上海市电站自动化技术重点实验室,上海200072 [2]新加坡国立大学工程学院,新加坡119260

出　　处：《上海交通大学学报》2016年第6期884-890,共7页Journal of Shanghai Jiaotong University

基　　金：国家自然科学基金项目(61374044);上海市科委国际合作项目(15510722100);上海市教委创新重点项目(14ZZ088);2013年上海市人才发展基金项目(201310)资助

摘　　要：提出一种在过程噪声干扰下基于偏差补偿递推最小二乘法的Hammerstein-Wiener神经模糊模型多信号源辨识方法.通过三阶段组合式多源信号,实现了过程噪声干扰下HammersteinWiener模型中输入静态非线性环节、动态线性环节以及输出静态非线性环节的分离,提高了各串联环节模型参数的分离精度,避免了传统迭代分离方法中模型参数的收敛问题.在此基础上,利用补偿原理,提出基于偏差补偿递推最小二乘法的Hammerstein-Wiener模型参数辨识方法,通过在递推最小二乘算法中引入修正项来补偿过程噪声引起的估计偏差,实现了模型参数的无偏估计.理论分析与仿真实验的结果表明,所提出的方法具有较好的有效性.Considering the process noise,a multi-signal identification of Hammerstein-Wiener models based on bias compensation recursive least squares was proposed.The fuzzy-neuro static nonlinear link and dynamic linear link of the Hammerstein-Wiener model with process noise were separated by the three-stage modular multi-signals,improving the separation accuracy of the link parameters of each series and avoiding the convergence problem of model parameters in the iterative separation method.Using the compensation principle,an identification of Hammerstein-Wiener models based on bias compensation recursive least squares was proposed to compensate the estimation bias caused by the process noise by introducing the correct item in the recursive least squares,and the unbiased estimation of the parameters was obtained.Theoretical analysis and computer simulation validated the effectiveness of the proposed algorithm.

关 键 词：HAMMERSTEIN-WIENER模型 信号源 偏差补偿递推最小二乘法 参数辨识 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]