Analog circuit soft-fault diagnosis based on sensitivity analysis with minimum fault number rule

Abstract

This paper proposes a new single or multiple soft analog circuit fault diagnosis approach based on the minimum fault number rule. It is based on the consideration that the fact that the probability of a single soft fault is much greater than that of a multiple fault if the related fault modes are independent. In this way, a new diagnostic strategy based on the circuit sensitivity analysis is proposed. The proposed strategy is an optimization-based one, whose objective is to find the minimum value of unaccepted parameter deviations which satisfy all those constraints, and the constraints equations are actually the voltage increment equations in all test nodes and the changing range of each element. The diagnosis process can fulfill the requirement of fault detection and fault isolation. It enables a fast or a real-time diagnosis in practical engineering. A DC circuit example and an AC circuit example are presented to demonstrate the effectiveness of the proposed approach.