Nonlinear Predictive Controller Design And Its Applications

To solve the problems which exist in chemical processes, such as, nonlinear, numerous variables, numerous constraints, uncertainty, and time-delay etc., model predictive control strategies are proposed for several kinds of nonlinear systems in this paper.Firstly, a nonlinear generalized predictive controller based on wavelet transformation is proposed. The predictive model of the nonlinear generalized predictive controller is Hammerstein model. The nonlinear static part of the Hammerstein model is identified by wavelet net, and the linear dynamic part is identified by least square method. The tracking properties of the closed loop system are verified by simulation for a sample continuous stirred tank reactor. The simulation shows that the closed loop system has well dynamic behavior.Secondly, a nonlinear model predictive controller for a class of single input single output affine nonlinear systems with uncertainty is designed. The controller expression with undetermined parameters can be acquired by using backstepping design idea, and the controller parameters are optimized online by using model predictive control. By using controller proposed in this paper, not only the stability of the closed loop system can be acquired more easily but also the dynamic behavior is good. The simulation of a continuous fermenter process also shows that the controller is valid.Thirdly, considering the time-delay problem which exists in many chemical processes, the nonlinear model predictive controllers for a class of single input single output time-delay nonlinear systems with system uncertainty is proposed. Based on an iterative procedure known as backstepping, the Lyapunov-Krasovskii functionals are constructed at each step. By magnifying inequality at each step, the expression of input can be acquired. If the time-delay is known, the stability of the closed loop systems can be guaranteed; if the time-delay is unknown, the uniformly ultimately boundedness of the closed loop systems can be guaranteed. By using the model predictive scheme, design parameters of a controller are on-line optimized, in this way the dynamic properties of the closed loop systems can be improved. An industry process - a two stage continuous stirred tank reactor - has been provided to illustrate the application of the main result. The simulation shows that the controller proposed in this paper has well controlled behavior.At last, a nonlinear model predictive control strategy via input output feedback linearization for multi input multi output systems when inputs and outputs constraints exist in the systems is proposed in this paper. The nonlinear predictive models are transformed into input output linear model by nonlinear coordinate transformations, and the constraints in new coordinate systems can be acquired by calculating optimized problems. The inputs in the new coordinate systems can be acquired by linear model predictive control arithmetic, and then the inputs of the nonlinear systems can be calculated by inverse transform of the nonlinear coordinates transform. Taking example for a complicated nonlinear chemical process, the fluid catalytic cracking unit, the feedback linearization controllers based on centralized parameter model and distributed parameter model respectively and the nonlinear predictive controller are designed. The simulation shows that the controller proposed in this paper has well controlled behavior.