Neural network-based event-triggered fault detection of discrete-time nonlinear uncertain systems

Abstract

This paper provides a novel neural network-based event-triggered fault detection (FD) strategy for nonlinear discrete-time systems with unknown function, unmodeled dynamics and disturbances. An FD observer is proposed in the framework of neural network (NN) approximation and event-triggered mechanism. Based on the stability analysis at the trigger instants and during the inter-event time intervals, the aperiodic NN weight update law is proposed to remain ultimately bounded of the state and weight estimation errors. The FD decision logic is proposed follows by fault detectability analysis. The scheme ensures the reduction of communication burden while maintaining FD performance. Simulation results state the validity of the method.