Decentralized Fault Detection in a Class of Large-Scale Nonlinear Uncertain Systems

This paper presents a decentralized fault detection scheme for a class of distributed large-scale nonlinear uncertain systems. The fault detection architecture consists of interacting local detection filters, one for each subsystem. Under certain assumptions, several important properties of the decentralized fault diagnosis scheme are rigorously investigated. The fault detection threshold for each decentralized detection filter is derived. The detectability analysis characterizes the class of faults that can be detected by the presented scheme. The fault detection time analysis provides an upper bound on the time it takes to detect a fault after its occurrence. The stability and performance analysis deals with the bondededness of all system signals in the adaptive estimation model and characterizes the bounds on the estimation error signals. A simulation example is used to illustrate the effectiveness of the proposed method.