Event-triggered fault detection for discrete-time linear multi-agent systems

Abstract

This paper studies the design and development of event-triggered fault detection (FD) filters for discrete-time linear multi-agent systems. For each agent, an FD filter is designed that receives the output measurements from its neighboring agents whenever specific event conditions are satisfied. With our proposed methodology all agents collaborate with one another to detect the occurrence of faults in the team and each agent not only can detect its own fault but also is capable of detecting its neighbor's fault. The filter parameters and the event conditions are designed such that a mixed H∞/H- performance index is guaranteed and it is shown that by using an event-triggered technique, the amount of data that is sent by each agent to its neighboring agents is dramatically decreased. Sufficient conditions for the solvability of the problem are obtained in terms of linear matrix inequalities (LMIs) where extended LMI characterizations are used to reduce the conservativeness of the multi-objective H∞/H- problem. Simulation results corresponding to a team of autonomous unmanned underwater vehicles demonstrate and illustrate the effectiveness and capabilities of the proposed methodology.