Research On Subspace Identification And Predictive Control Methods For Wiener Nonlinear Systems

Wiener non-linear systems are a typical class of modular non-linear systems,the accurate identification and "stable,accurate and fast" control of which is of great importance for chemical production processes.As a powerful tool for identifying state space models,the subspace identification method has the advantages of simple numerical calculation,strong numerical robustness and no non-linear optimisation process.In this paper,a new method is proposed for the identification of Wiener nonlinear systems,which is difficult and has over-parameterisation problems;and the existing predictive control methods for Wiener systems need to be implemented with the knowledge of the parametric model of the system,which has problems such as complex calculations and large computational effort.In this paper,a new subspace identification method for Wiener nonlinear systems and a new predictive control method for Wiener nonlinear systems incorporating data-driven ideas are proposed,and the main work is as follows:1.A new method for subspace identification of Wiener systems based on hierarchical iterations is proposed to avoid estimating unnecessary redundant parameters and to improve the identification accuracy.Firstly,the predictor model is introduced through the problem description of Wiener system subspace identification,and the key term separation technique is used to divide the dynamic linear and static non-linear into two sub-models.Then a two-stage hierarchical iterative least squares method is proposed to estimate the unknown parameters of the two sub-models separately.2.A data-driven predictive control method for Wiener systems based on the subspace identification algorithm is proposed,which implements a control strategy for designing controllers directly from data and reduces the computational effort of the predictive control process.The subspace predictor model is built from the collected input and output measurement data,enabling the calculation of the subspace predictor parameter matrix only to predict the output of the system when the parameterised model of the system is unknown.Subsequently,a data-driven predictive controller is designed using the subspace predictive output.