Application Research Of Predictive Control Based On RBF-ARX Model On Water Tank Level System

Currently, the progress of non-linear system modeling and control research is not so perfect that restricts the further development of industrial production. Therefore, research on the applications of prediction control based on RBF-ARX model has positive sense for the development of control theory and industrial production. The water tank level control system has the same characteristics with the complicated industrial process control systems. RBF-ARX Modeling and model predictive control (MPC) to the water tank level control system will be presented in this thesis, which may include following aspects:Firstly, for MATLAB’s strong ability of calculation, the DDE communication link is built between MATLAB and Kingview, while Simulink clock synchronization is designed. Thus, a real-time control platform for water tank level control system is exploited in this thesis, which is combined with the advantages of MATLAB and Kingview, with a simple configuration and strong ability of calculation.Secondly, the analysis of nonlinear and coupling of the water tank is done by means of physical model and the actual sampling data. Using the global and local characteristics of RBF-ARX, the thesis describes the dynamic relationship between upper and lower water tank level and the opening degree of electric control valve by MIMO-RBF-ARX. A two input and two output RBF-ARX model derived from a wide range of sampling data of level system whose effectiveness has been proved by testing data. Then, this model is changed into state equation in order to design a MPC controller based on MIMO-RBF-ARX and the controller is implemented on the simulink platform with M language, and the simulation has finished. The simulation results show that the MPC algorithm based on RBF-ARX model succeeds in precise control of water tank level.Finally, the real-time control of the MPC algorithm based on RBF-ARX model is done on actual object. The comparative results between the application of PID control and the RBF-ARX model-based predictive control to the real water tank level control system are displayed in this thesis, it shows that the predictive controller based on RBF-ARX model has better effects.