Finite-time Fault Estimation Based on Unknown Input Observer for Multi-agent Systems

In this paper, we introduce a finite-time fault estimator capable of estimating both constant and time-varying faults for multi-agent systems with a leader-follower structure. Based on the relative output information of each agent, each subsystem is transformed into a large integrated system by Kronecker product theory. By using the spectral decomposition and coordinate transformation, the integrated system is decoupled into independent subsystems. For each subsystem, two unknown input observers with different parameters are designed. It can be proved that the fault estimation error converges to zero within any given time delay. In addition, the external interference is completely decoupled. At the end of this paper, a numerical simulation is presented to validate the authenticity and effectiveness of the proposed method.