Composite Neural Learning Fault-Tolerant Control for Underactuated Vehicles With Event-Triggered Input

This article presents a novel composite neural learning fault-tolerant algorithm to implement the path-following activity of underactuated vehicles with event-triggered input. With the input event-triggered mechanism, the dominant superiority is to reduce the communication burden in the channel from the controller to actuators. In the proposed scheme, the system uncertainties are dealt with in the fusion of the neural networks (NNs) and the dynamic surface control (DSC) method. The serial-parallel estimation model (SPEM) is constructed to estimate the error dynamics, where the derived prediction error could improve the compensation effect of the NNs. As for the gain uncertainties and the unknown actuator faults, four adaptive parameters are designed to stabilize the related perturbation and not be affected by the triggering instants. Based on the direct Lyapunov theorem, considerable efforts have been made to guarantee the semiglobal uniformly ultimately bounded (SGUUB) stability of the closed-loop system. Finally, comparison and practical experiments are illustrated to verify the superiority of the proposed algorithm.

Automated summary

This article presents a novel composite neural learning fault-tolerant algorithm for implementing path-following activities of underactuated vehicles with event-triggered input. The algorithm reduces communication burden and handles system uncertainties through fusion of neural networks and dynamic surface control. By utilizing adaptive parameters, gain uncertainties and unknown actuator faults are stabilized to ensure the stability of the closed-loop system. Practical experiments demonstrate the superiority of this proposed algorithm.