Application Of Nonlinear System Modeling And Predictive Control On Stabilization Control Of Inverted Pendulum

The inverted pendulum system is always a hotspot in control field in recent years, which is a multivariable, nonlinear and strong-coupling natural instability system. The research of the inverted pendulum has very important value, not only the theory but also the realistic meaning in the research such as the walking of biped robot, the lunching process of rocket and flying control of helicopter. Based on the physical model of inverted pendulum system, the characteristics of the inverted pendulum system were analyzed. Some modeling and control strategies of inverted pendulum were compared. According to the non-linear characteristics of the inverted pendulum system, predictive control based on RBF-ARX would be used to control the inverted pendulum system.Usually, most of the researches on the inverted pendulum were based on the traditional physical methods. The physical models were established based on the differential equations deduced by mechanism analysis. It has many inevitable defects. For example, the physical parameters are difficult to be precisely measured, and some simplifications and linearizations are made in the process of modeling, All of these would degrade the accuracy of the model.To overcome these disadvantages of the physical method, the identification method is proposed and applied on the inverted pendulum. We choose RBF-ARX model to do the modeling work of inverted pendulum, which has an excellent performance in approximation in a linear space and can be used for model predictive control easily. In this paper, the structure and principle of the RBF-ARX model was introduced, and the identification, optimization and selection of model’s parameters were analyzed and explained in detail. The data which can describe the dynamic characteristics of inverted pendulum was sampled. The RBF-ARX model of the inverted pendulum system was built after estimating the parameters offline by SNPOM. At last, the RBF-ARX model and the ARX model with the same order were compared. The results showed that the RBF-ARX model had higher modeling accuracy. Based on the RBF-ARX model, we designed two predictive controllers separately for the SISO system and the SIMO system, and obtained good simulation results under different intial conditions, meanwhile, we also selected the best controller parameters. After that, the RBF-ARX model-based predictive control was carried out on the real inverted pendulum system, the application on the SISO system was susscessful. The simulation and real-time control results make sure that MPC is relisbly and the RBF-ARX model is effectively.