The Research On The Application Of Inverted Pendulum System With RBF Neural Network Based On Backstepping Method

The inverted pendulum is a typical model in the cybernetics. Because it is full of unstable, nonlinear, multivariable, strong-coupling characteristics, and its control process can effectively reflect many significant control problems, such as nonlinear problem and robust problem. The inverted pendulum system is often used to testify the effectiveness of a new control theory and control method in the laboratory. Also, the control of the inverted pendulum system is similar with the control of the engineering, such as robotics well, so the examination of the new control methods in the inverted pendulum has a good guide to the other Engineering control.The paper researches the control of linear inverted pendulum system based on the Radial Basis Function Neural Network (RBFNN) controller which is designed by the use of Backstepping method with RBFNN. Backstepping method can decompose the complex nonlinear system into several low-level sub-systems, and the Lyapunov function and inter virtual control variables of every subsystem are derived. The design of the controller is completed based on Backstepping method for the nonlinear systems. The main works of this article are:1. The inverted pendulum system is discussed and analyzed, and Newton mechanics and Lagrange method are adopted respectively to establish mathematical models, and therefore their equations of state are deduced.2. Based on the discussion of Backstepping method and RBF Neural Network, a controller which combined Backstepping method with RBFNN is formed, and the controller is applied to linear inverted pendulum system. The results indicate that our method is feasible. 3. According to the content which is described above, the module of RBF Neural Network controller is built by the simulink toolbox of the MATLAB, and simulation is conducted by linear inverted pendulum system.The experimental results show that the method which combines Backstepping method with RBF Neural Network is active and effective in controlling the inverted pendulum.