Anti-disturbance Identification Of Linear And Hammerstein Nonlinear Subspace Models

System identification is a commonly used mathematical modeling method.However,in actual industrial processes,load disturbances are always present and have an impact on the quality of the product.Most of the existing algorithms do not consider the load disturbance factors,which leads to inaccurate estimation of the model parameters and affects the controller design,so the control effect is not ideal.Therefore,it is of theoretical significance and practical guiding value to study the parameter estimation problem under load disturbance.In this thesis,the corresponding subspace identification method is proposed for linear systems and Hammerstein nonlinear systems under load disturbance.The research content can be summarized into three levels: from the view of the two methods of dealing with load disturbance,it is divided into "elimination" and "variation";for the system classification,the linear system and Hammerstein nonlinear system are explored;for the data identification method,this thesis uses online and offline algorithms.The main work is summarized as follows:Firstly,aiming at the linear system immunity problem in industrial process,the influence of known types of load disturbance on linear system identification is explored,and based on the dynamic characteristics of known types of disturbances,a system based decomposition and LQ decomposition is proposed.Through the system model decomposition,the system is decomposed into three parts,including the determination part,the random part and the disturbance part.The decomposed system is augmented,then QR decomposition is performed to eliminate the influence of load disturbance,and then the SVD decomposition is used to extract the system matrix to be identified.This method can eliminate the influence of disturbance on the linear system through LQ decomposition and achieve good identification.The proposed method can not only be applied to open-loop linear systems,but also avoid the feedback coupling problem caused by closed-loop system feedback controllers,so that the subspace anti-interference identification of closed-loop linear systems can also be applied.At the same time,through the MATLAB simulation example,and comparing with the existing subspace identification algorithm,the superiority and practicability of the proposed algorithm are highlighted.Secondly,for the Hammerstein nonlinear system with slow time-varying load disturbance under the state space model,the slow time-varying disturbance is treated as a parameter to be identified.Based on the separation strategy,the system is decomposed into two parts.The first part is the corresponding state space output,and the two parts are about the time-varying load disturbance response.The least squares algorithm is constructed for these two parts,and the adaptive forgetting factor is introduced.The identification results of the two parts are obtained.Since the real output of the system in the information vector is unmeasurable,the real output is inferred by predicting the normal output.At the same time,the effectiveness and superiority of the proposed algorithm are proved by comparing the simulation examples with other algorithms.Finally,the convergence analysis of the proposed algorithm is based on the continuous excitation condition.