A Class Of Nonlinear System Identification Based On Combined Signal Source

The identification of nonlinear systems is of great significance in actual life.In system identification,a popular research direction is the system identification of a typical nonlinear block structure model.In the typical nonlinear block structure model,through the integration of dynamic linear module and static nonlinear module,the process control characteristics are well reflected,which makes it easier to identify,the amount of calculation is reduced obviously,and the application is more extensive.According to the connection mode,it is divided into Hammerstein model,Hammerstein-Wiener model and so on.Hammerstein model and Hammerstein-Wiener model are mainly identified in this paper.The main tasks are as follows:1.According to the state space model of Hammerstein system,the binary pseudorandom inverse M sequence coded signal and random coded signal are fused to form a combined signal source to decouple the static nonlinear module and the dynamic linear module.By using the neural fuzzy network model,the static nonlinear module can be fitted with high precision,which can effectively avoid the limitation that the traditional Hammerstein model uses polynomial to approach the nonlinear function,thus broadening the applicable scope of the identifiable nonlinear model.By using the combined signal source,the state space equation model identification technology of nonlinear module and linear module can be realized,and then the subspace identification technology can be applied to the state space equation model,which can reduce the amount of calculation and improve the operation efficiency.The system identification of the nonlinear saturated dead zone model and a more complex nonlinear model are carried out respectively,and the simulation results show that the proposed algorithm is feasible and effective.2.For the Hammerstein-Wiener system,the state space model is used in the linear module.To reduce the iterative times of solving the model parameters and solve the convergence problem of the model parameters,the binary pseudorandom inverse M sequence coded signal and the random coded signal are combined to separate the static nonlinear module from the dynamic linear module.on this basis,a neuro-fuzzy Hammerstein-Wiener model parameter algorithm based on Taylor series expansion is proposed.The algorithm further increases the ability to approach nonlinear modules.The effectiveness of the algorithm is verified by a simulation example.3.Aiming at the example of p H neutralization process,the corresponding state space model of Hammerstein system is given.The subspace identification method based on the combined signal source is used to simulate and verify the given model,which further verifies the effectiveness of the subspace identification method based on the combined signal source.In this paper,the simulation is carried out according to the proposed identification method,and the research work is summarized as well.Finally,the main researches and work directions in the future are sorted out,and the prospect is made in the end.