Communication-based Fault Detection and Isolation for Multi-agent Systems

In this paper, a communication-based fault detection and isolation strategy for multi-agent systems is proposed and all the fault processing operations are achieved by controlling the contents of communication. A fault detection scheme based on the gossip algorithm and status information exchange is introduced first. Then, a new fault isolation and recovery algorithm is designed to moderate its negative effects on control protocols. Considering the topology destruction the strategy may cause, a virtual path, by using two-hop transfer information, is created to confirm that the system can keep connectivity at all time and the normal process will not be influenced by the fault processing operation. All the schemes are proved to be effective in theory and several simulations are presented to demonstrate the efficiency of the strategy.