Recursive Identification For Hammerstein-Wiener Nonlinear Systems With Non-uniform Sampling

The Hammerstein-Wiener nonlinear systems consist of a static input nonlinear subsystem,a dynamic linear subsystem and a static output nonlinear subsystem.The Hammerstein-Wiener systems have attracted numerous attention of researchers owing to their effectual reflections to the structure of typical nonlinear systems.In order to research the Hammerstein-Wiener system,the parameters of the system must be obtained.Therefore,it is of great theoretical significance and application prospects to study the parameter identification methods of the Hammerstein-Wiener system.In the actual industrial process,the sampled data may be in a non-uniform state due to the limitations of hardware and environment,so the study of the non-uniform sampling system is of general practical value.This paper considers the Hammerstein-Wiener nonlinear system with non-uniform sampling,the recursive parameter identification methods are studied by using the observation data.The main contents are as follows:1.For the Hammerstein-Wiener nonlinear system with non-uniform sampling,the overparameterization identification model of the system is derived.The parameter vector of the over-parameterization model contains the parameter product terms of the input nonlinear part and the linear subsystem.By means of the least squares principle and the gradient descent principle,this paper considers these product terms as independent terms and proposes a class of least squares algorithms and a class of stochastic gradient algorithms based on the over-parameterization identification model.Finally,the average value method is used to separate the real parameters to be identified from the independent terms,and then the system parameter estimates are obtained.2.Due to the over-parameterization identification model contains the parameter product terms,the key-term separation identification model of the system is derived by using the key-term separation technique.Compared with the over-parameterization identification model,the parameter vector of the key-term separation identification model does not contain the parameter product terms.For the unknown internal variables in the key-term separation identification model,in terms of the auxiliary model identification idea,the unknown internal variables are replaced by their estimates and a class of least squares algorithms and a class of stochastic gradient algorithms are proposed based on the key-term separation identification model.In conclusion,this paper comprehensively studies the parameter identification problem for the over-parameterization identification model and the key-term separation identification model of the Hammerstein-Wiener nonlinear system with non-uniform sampling.The identification steps of each parameter identification algorithm and the corresponding simulation examples are given,and the effectiveness of the parameter identification algorithm proposed in this paper is verified according to the simulation results.Finally,the thesis draws conclusions and prospects,and the problems to be solved in the identification of the Hammerstein-Wiener nonlinear system and other aspects worthy of further study are briefly introduced.