Search Of RBF-ARX Model-based Predictive Control On Water Tank Level System

Progress on the development of industrial production plays an essential role on the national economy. With industrial process increasing more and more complexly, the study of advanced control algorithms applied in industrial process is important as a result of the arising of traditional PID control’s shortcomings. Water tank level system is a coupled and time-delay nonlinear system including common industrial process controlled parameters such as flow, level, temperature and pressure. Study about the modeling of water tank level system and predictive control has been done in this thesis, which will be meaningful to industrial process control.1. In order to monitor the real-time control system and collect the experiment data, the communication between water tank equipment with Kingview, and the DDE communication between Kingview with Matlab software are built in this thesis, and Matlab is used to write control algorithms for experimental research.2. Based on the actual data which is collected by PID controller, ARX model and RBF-ARX model both belonging to identification model are acquired respectively by using the LSM and SNPOM to estimate the parameters. And the compared results confirm that the RBF-ARX modeling approach more suitable for modeling of the water tank level system.3. Predictive controller is designed based on the RBF-ARX model that is used as the predictive model, and simulation platform is built by using Matlab for RBF-ARX-MPC experiment for the deep understanding of the parameters effects. The best parameters are got through several simulation experiments, which provide a better adjustment direction for the real-time experiment.4. The real-time experiment are done by using PID control algorithm and RBF-ARX model-based predictive control algorithm respectively in this thesis, and the comparison between the two kinds of results show that predictive control algorithm based on RBF-ARX model has achieved fast response, small overshoot and is more suitable for the control of the nonlinear system.