Fault detection and isolation using relative information for multi-agent systems

We study fault detection and isolation of multi-agent systems existing fault signals. Based upon the designing idea of unknown input observer, we proposed an innovative fault detection and isolation strategy. When the existence conditions are satisfied, the observer can be constructed using only relative information. A threshold logic and the corresponding algorithm are presented to detect and isolate the agent suffering from fault. Furthermore, a distributed implementation of our method is provided which helps to reduce the observer dimensions and simplify the design procedure. Practical simulations are given to demonstrate the validity of the theoretic results.