Based On The Intelligent Sliding Mode Variable Structure Control Of Magnetic Levitation Ball System Research

As a key object of magnetic levitation technology research, the magnetic levitation ball system is a typical nonlinear and unstable system. The research on the system involves electromagnetism, control theory, digital signal processing science and computer technology.Because of its high real-time requirements, using traditional control methods is difficult to control the system. On the basis of the relevant research work both in abroad and at home, looking for the control method to improve the overall performance of the magnetic levitation ball system has important significance in research.This paper has several parts. Firstly, it introduced the magnetic levitation technology research and development overview. According to the composition and working principle of the magnetic levitation ball system,described its mathematics model,then analyzed the linearization characteristics of the nonlinear mathematical model nearby the balance point. Then on the basis, it built a classic sliding mode controller to control Mag Lev ball system in MATLAB.Secondly, it designed a second-order dynamic sliding mode controller to solve the inevitable “chattering” problem in switch surface.Instead, introduced a new sliding mode switching surface which contained the higher derivative of inputs, reducing the impact of discrete item in reaching law could eliminate the “chattering” effect. Then, a new sliding mode equivalent control based on RBF neural network is proposed by combining the features of variable structure and RBF neural network. The advantage of this control method is making full use of neural network’s self-learning and self-processing abilities to improve the control effect in the system. In order to verify the control effect of the controller, built the control model in MATLAB and simulation.Finally, it introduced a Quanser magnetic levitation ball system semi-physical simulation experiment platform, applied the designed control method in hardware in loop device for real-time control experience. The experiment proves that designed methods have better control effect and stronger robustness.