ADP-based event-sampled optimized consensus tracking of stochastic multi-agent systems with actuator faults

Abstract

This article concentrates on the event-triggered optimized leader-follower consensus control for stochastic nonlinear multi-agent systems (MASs) with actuator bias faults. An adaptive state identifier is designed to estimate unknown states in real time. In addition, an adaptive dynamic programming (ADP)-based algorithm with a critic-actor architecture is developed to learn the optimized control policy online. Based on the online estimating information from the designed fault estimator, a fault-tolerant (FT) controller is designed to compensate for the deleterious effects of actuator bias faults. The proposed control scheme employs a dynamic event-triggered mechanism (ETM) to reduce network resources by tuning dynamic parameters. All the signals can be proven to be bounded by utilizing the Lyapunov stability method, and the Zeno behavior can be excluded. Simulation results show the merit and validity of the proposed control scheme.