Fully Distributed Cooperative Fault-Tolerant Output Regulation of Linear Heterogeneous MASs: A Dynamic Memory Event-Triggered Distributed Observer Approach

This paper presents a fully distributed, reliable output regulation scheme on the basis of a dynamic memory event-triggered mechanism for a class of linear heterogeneous multiagent systems under actuator faults. First, dynamic memory event-triggered mechanisms are established by introducing historical information on dynamic variables. Rigorous analysis shows that the developed dynamic memory event-triggered mechanisms produce larger event intervals than the existing memoryless dynamic event-triggered mechanism. Furthermore, in cases where only a portion of the followers possess knowledge about the exosystem matrix, an open-loop estimation approach is devised to circumvent the need for ongoing communication among the followers. In contrast to the current approach for intermittent communication through open-loop estimation, the proposed approach removes the need for all agents to possess identical and pre-known system matrices. Additionally, the proposed scheme enables all followers to obtain cooperative fault-tolerant output regulation fully distributed while avoiding Zeno behavior. Finally, the efficacy of the developed strategy is substantiated by simulation results. Note to Practitioners—This study focuses on the issue of multiagent cooperative output regulation, which has practical applications in various systems, including robot formation and satellite formation. In such scenarios, the control objective can be transformed into the issue of achieving cooperative output regulation in multiagent systems. Unlike the existing works, this study introduces a novel dynamic memory event-triggered mechanism aimed at enhancing the minimum event-triggered intervals. In addition, actuator faults are prevalent in engineering. The developed scheme ensures the reliable operation of the controlled system, with minimum event-triggered intervals larger than small positive constants. Constant communication between agents is avoided, which is beneficial for application in practical systems. Ultimately, the effectiveness of the scheme is verified through applications in mobile robot formation and satellite formation. The proposed method is anticipated to be expandable to multiagent systems characterized by a wider range of practical limitations and to be utilized in more realistic practical engineering.