RBF-ARX Model-based Predictive Control With Application To Magnetic Levitation System

The magnetic levitation technology is an attractive field of the current control research domain. It is the key technology of magnetic levitation train, magnetic levitation balance, magnetic bearing and magnetic motor and so on. The Magnetic Levitation Control System is the platform of magnetic levitation technology studies, due to the characteristics of essential nonlinear, uncertainty and open loop instability, so it is hard to establish the precise mathematical model. The RBF-ARX model is a modelling method which aims at the complex nonlinear industrial process control. The RBF-ARX model is the composite construction of Radial Basis Function (RBF) network and the linear AutoRegressive model with exogenous (ARX), it is a global model and able to describe the system behavior. RBF-ARX model is estimated off-line by SNPOM which is a fast-converging hybrid method composed of Levenberg-Marquardt method (LMM) and least-squares method (LSM). Local linearization of the system at each working-point may be easily obtained from the global RBF-ARX model. The predictive control based on RBF-ARX model is a new forward question in the domestic and foreign nonlinear industrial process control. According to the basic principle of the predictive control, this thesis discussed the RBF-ARX model structure and the predictive control based on RBF-ARX (RBF-ARX-MPC). Moreover, it also discoveried the application of predictive control based on RBF-ARX model in The Magnetic Levitation Control System.Firstly, this thesis introduced the present research and application of the magnetic levitation technology in the domestic and foreign, and established The Magnetic Levitation Control System's mathematical model according to the system's each constituent's physical principle, and obtained system's state equation and the transfer function through to carring on the linearization of the nonlinear mathematical model nearby the balance point. Then we analyze the model characteristics of controllability, open loop stability.Secondly, this thesis introduced predictive control's development process and the questiones which exist in its application, and specifically introduced predictive control's basic principle, and discussed the predictive control basic structure take the generalized predictive control algorithm as the foundation, analyzed its predictive model, methods of revising feedback and receding horizon optimization.Thirdly, this thesis emphasisly introduced the RBF-ARX model structure, the principle and method about parameter identification. Then it established The Magnetic Levitation Control System's RBF-ARX model under different state variable, and analyzed each kind of model reliability.Finally, this thesis introduced the principle of predictive control algorithm which is based on the RBF-ARX model, and carried on the control simulation to The Magnetic Levitation Control System through RBF-ARX model, simultaneously, discoveried the certain feasibility when we carried on the real time control. At last, we used PID controller and Forward-Backword controller to The Magnetic Levitation Control System and geted good control performance.