Study On Magnetic Levitation System Base On Fuzzy RBF Neural Network

Magnetic levitation is typical technologies which are combined with electromagnetism, dynamics, power electronics and automation and develop with these technologies and theory. In nowadays, more and more domestic and foreign scholars have devoted themselves to the study of magnetic levitation technology. Magnetic levitation system is complex systems which are nonlinear; open-loop unstable and uncertain, these characteristics adds the difficulty of designing the maglev controller. The conventional PID control is difficult to obtain satisfactory control effect. A new control strategy must be researched and developed.In this paper, firstly to make single degree of freedom magnetic levitation system as the research object and deeply study its structure, operating principle and dynamic performance. According to the characteristics of the system such as open-loop instability and strong nonlinearity, the nonlinearity model of maglev system is programmed by S-function, then control the magnetic levitation system separately by the conventional PID control and BP neural network PID control and simulate them in MATLAB environment, confirm the flaws and shortcomings by comparing the results of these two control methods. Finally, to solve the problem of the parameters adjusting of conventional PID control being difficult and BP algorithm of neural network training easily falling into local minimum point as well as the low learning efficiency, a kind of intelligent PID control system based on fuzzy RBF neural network method is proposed. Radial basis function (RBF) neural network has the advantage of small computation and fast convergence, can improve the features of BP network. This method utilized with the good learning ability of neural network and strong reasoning ability of fuzzy control to adjust the three parameters kp,ki and kd of PID control online by combining the fuzzy reasoning and RBF neural network. Adjust set suitable parameters for maglev system to meet the requirements of maglev systems static and dynamic performance. The simulation results shows that the intelligent PID control system based on fuzzy RBF neural network method has better adaptability and robustness which can control magnetic levitation system more effectively.