A hybrid stochastic-deterministic input design method for active fault diagnosis

Abstract

The problem of designing an input to improve the detectability of faults has been addressed previously using both stochastic and deterministic formulations. This article presents a hybrid approach that provides a worst-case guarantee of diagnosis within a time interval [0,N], while maximizing the probability of diagnosis at some earlier time M <; N. Compared to purely deterministic methods, this strategy reduces the average time required for diagnosis. Moreover, a high probability of diagnosis at M enables N to be chosen large, thus reducing the conservatism of the required input.