Model-Based Control

This is a doctoral thesis about prediction and control. It summarized basic situation of predictive control, and narrated a few typical control algorithms of predictive control, and discussed the problems of control models, closed-loop characteristics, parameter design. To developing nonlinear predictive control and the predictive control based on artificial neural networks, it made researches.In control algorithms, this thesis selected a few of algorithms with typical meaning, and deduced their control laws, showed algorithm steps. Relating a liquid level container, it introduced algrothm useage and simulation. This thesis led self-tuning techniqueint to dynamic matrix, and made the system's adaptation and robusness enhanced. Algorithms' formulation is clear, and convenient to programming and calculation.In accordance with multi-models used in predictive control, this book revealed the relation between different models, and offered their conversion formulas, which provides convenience to the analysis and design of predictive control. Through the analysis of the closed-loop characteristics of predictive control systems, it showed that internal model control is a powerful tool to various predictive control systems. Based on the conclusions of stability and robustness of predictive control systems, this thesis qualitatively expounded the relation between design parameters and control characteristics, and verified the conclusions with an example, which has a guide meaning to the design of predictive control.In nonlinear control, this book mentioned 3 methods: approximation method, Hammerstein model method, and the method of nonlinear autoregressive with exogenous input (called NARX for short). The problems caused by extracting a root with the approximation method based on Hammerstein model were studied in depth. Some sure conclusions were gained. It was considened a good method for handling nonlinear, approximating nonlinear with NARX model, and simplifing the model...