Neural adaptive coordination control of multiple trains under bidirectional communication topology

This paper investigates the problem of coordination control for a group of trains by neural adaptive approach. The communication structure among trains is a bidirectional one, i.e., necessary information of neighboring trains is used in the control design for a train. Two control schemes are developed, with the first one requiring the information of position, speed, and acceleration of neighboring trains, while the second requiring the information of position of neighboring trains only by virtue of high-order sliding mode observer technique. Based on the universal approximation capacity of radial basis function neural networks, there are no requirements of the precise parameters describing operational resistance and other kinds of extra resistances in the controller design, which are reconstructed by radial basis function neural networks online. The stability of single train and multiple trains are guaranteed by Lyapunov stability theorem. Numerical simulations are presented to demonstrate the effectiveness and performance of the proposed controllers.