Performance-Guaranteed Multi-Train Cooperative Control using Event-Triggered and Sign-Rectified Regressor

This paper presents a performance-guaranteed method of cooperative tracking for multiple high-speed trains while considering computational resource limitations and security constraints. First, the prescribed performance control is employed to transform constrained train states into an unconstrained format. Radial basis function neural networks are used to efficiently approximate the unknown resistance, and an innovative sign-rectified regressor neural network update strategy is employed to significantly improve computational efficiency and ensure high estimation accuracy even with limited neuron number and weak excitation. Secondly, an event-triggered mechanism is designed to determine the timings of updates for the train controller and neural network, adapted for on-board computational resource limits. Subsequently, the triggering conditions for each train are then designed to facilitate a safe cooperative control method, which combines an event-triggered processor with a performance control protocol that ensures limited state tracking error within a predefined performance function. Finally, in order to validate the effectiveness of the proposed scheme, we applied the control strategy in simulation experiments using collected data from the high-speed railway line.