基于最小Wilcoxon学习方法的Hammerstein模型辨识

Identification of Hammerstein Model based on Least Wilcoxon Learning Method

提出一种基于最小Wilcoxon学习方法的非线性动态系统建模方法。用非线性静态子环节和线性动态子环节串联——Hammerstein模型来描述非线性动态系统。然后,将Hammerstein模型的非线性传递函数转换为等价的线性形式,从而建立起线性中间模型。再由最小Wilcoxon学习方法辨识出中间模型参数。最后,通过中间模型参数与Hammerstein模型参数之间的关系,推出原系统的非线性静态环节和线性动态环节的参数,从而实现原非线性动态系统建模。在系统仿真响应信号有扰动时,该方法比用最小二乘法辨识中间模型表现出更强的鲁棒性。

A modeling method for nonlinear dynamic system based on Least Wilcoxn learning method （ LW） was proposed .The nonlinear dynamic system is discribed by Hammerstein model expressed by a nonlinear static sub -unit followed by a linear dynamic subunit .Through the function expansion , the nonlinear transfer function of Ham-merstein model is converted to the equivalent linear form to generate the intermediate linear model .By LW meth-od, coefficients of the intermediate model are obtained .According to the relations between coefficients of interme-diate model and that of Hammerstein model , parameters of the nonlinear static subunit and linear dynamic subunit are derived .The original nonlinear dynamic system is modeled .The proposed method has better robustness than least square （ LS） method to identify the intermediate model , when there exists disturbance in output signal of the system.