抄録
This paper presents an adaptive nonlinear control approach to a 4-point attraction magnetic levitation system using neural network. From the electromagnetic theory, we know the electromagnetic model of the magnetic levitation system is usually strongly nonlinear where coefficients change depending on the length of the air-gap, and the system can be represented in the normal form of a 4-input/4-output system based on some assumptions, where the nonlinear functions in the system are unknown. In this paper, we present a hybrid neural network using the generalized radial basis functions to model the nonlinear functions. The advantage of the hybrid neural network is that it is easy to choose the initial parameters based on the nominal values of the system paremeters, and the adaptive nonlinear controller can be designed to control the system to track a reference input. The parameters of the neural network are updated in an on-line manner according to an augmented tracking error. The results of the convergence analysis of the adaptive nonlinear control system are also shown. At last, experimental results are included to show the excellent performance of the designed adaptive nonlinear control system.
