Neural Network-Based Adaptive Control for EMS Type Maglev Vehicle Systems with Time-Varying Mass

Abstract

The Electromagnetic Suspension (EMS) type maglev vehicles suffer from various control complexities such as strong nonlinearity, open-loop instability, parameter perturbations, which make the control of magnetic levitation system (MLS) very challenging. In actual operation, ever-changing passengers will cause the mass of vehicle body to deviate from the nominal mass of the levitation controller design, which will deteriorate the controller performance significantly. Therefore, the passenger carrying capacity of the maglev vehicle is strictly restricted, which restricts the further promotion of the maglev traffic. In this paper, an airgap control strategy based on the neural network (NN) is proposed for maglev vehicles with time-varying mass. Specifically, firstly, a nonlinear controller is proposed to guarantee the asymptotic stability of the closed-loop system. Next, to tackle the unknown or time-varying mass of vehicle body, an adaptive radial basis function (RBF) NN is utilized together with the proposed nonlinear controller. The stability analysis of the overall system is provided by Lyapunov techniques without any linearization to the original nonlinear model. Finally, a series of simulation results for the maglev vehicle are included to show the feasibility and superiority of the proposed control approach.