The RBF-ARX Model-based Nonlinear System Modeling And Predictive Control To Magnetic Levitation System

While urbanization brings ever more subways under construction, the enormous costs yet pose a problem for policy makers. Magnetic levitation train, however, might serve as an adequate substitute. Likewise, such technology can be equally applied to magnetic levitation balance, magnetic bearing, magnetic motor etc. As it is, non-linearity and open-loop instability make the magnetic levitation system less controllable. Therefore, to devise a way to keep it on a tight rein is essential.Experiments are carried out on a lab-installed magnetic levitation equipment, whose electromagnetic force sustains a hovering steel ball. Begun with an overview on the research and application of magnetic levitation, this paper presents in sequence the rudiments concerning physical model, RBF-ARX model, LQR theory and predictive control theory. The chapters that follow deal mainly with the global dynamic input-output data acquired through a PID controller. These data, after processed, yield the RBF-ARX model of the system. further more, the real-time control is realized by a LQR controller. And according to the state-space of the RBF-ARX model, we design the predictive control strategy and simulate the entire system to come to a conclusion.In this paper, we demonstrate at length how to select and optimize the parameters involved. Some of them are obtained by theory, and the rest by repeatedly debugging. One way or the other, we expect to improve the accuracy of the real-time control and the simulation. As are revealed, the RBF-ARX model excels in describing the dynamic magnetic levitation system; and LQR controller based on the model turns out to be a distinguished real-time controller, which works just as well in stationary state as in dynamic condition; and above all, based on the RBF-ARX model, the predictive controller functions satisfactorily in simulation. Hopefully this paper will generate more public discussion on industrial control process, and this outstanding model will be applied in more and more industrial applications.