Identification Method Of Wiener Multivariable System

The Wiener model is composed of a dynamic linear link and a static nonlinear link.,it is a modular nonlinear system.The modular nonlinear system is studied in this paper.The basic least square identification algorithm is introduced;Based on the separation of key variables and the idea of matrix transformation,the recursive least square method for Wiener model is proposed;Based on matrix transformation technology and multi innovation identification principle,a multi innovation stochastic gradient identification algorithm for Wiener model is proposed.Finally,Matlab simulation software is applied to simulate the above two identification algorithms.The main research work of this paper is as follows:1.Introduces the modular nonlinear system;The basic steps of nonlinear system identification are introduced;Several different types of Wiener models are introduced respectively.2.Introduces the basic principle of a least square identification algorithm,then introduce recursive thought,the recursive least squares identification principle is introduced,which has the characteristics of small computation and fast operation speed.In addition,we introduce multiple innovation identification method,stochastic gradient identification method and multi innovation stochastic gradient identification method.3.According to the Wiener multivariable model,the system is simplified and decomposed through key variable separation technology and matrix separation technology.The decomposed subsystem is a linear autoregressive model.Therefore,the recursive least square identification method can be used to estimate directly.4.According to the Wiener multivariable model,the system is decomposed by introducing the switch function to express the nonlinear link and the matrix separation technique.The obtained subsystem is represented as the autoregressive model,and the multi innovation stochastic gradient identification method is carried out to estimation parameters.5.The recursive least squares identification method and multi innovation stochastic gradient identification method are simulated respectively to verify the accuracy and effectiveness.