An unknown input observer based approach to fault detection for linear multi-agent systems

This thesis deals with the problem of fault detection and isolation (FDI) in multi-agent systems (MAS) with linear dynamics. After a brief introduction of MAS, the relative problems and applications, the problem of FDI is introduced. A model in which two kinds of faults (one on the actuator and one on the sensor) is presented, and an unknown input observers (UIO) technique is used in order to generate the residual signal that is necessary to detect the presence of a fault. Subsequently, the consensus problem is solved simultaneously with the FDI problem. The thesis investigates both the scenario in which the connection topology among the agents is directed and when it is undirected. Finally, some simulations with MATLAB are performed in order to shown the effectiveness of the proposed approach.