A Method for Quantitative Fault Diagnosability Analysis of Systems with Probabilistic Sensor Faults

摘要

Quantitative analysis of diagnosability performance provides a deep insight into the structural property of systems, which enables the selection of a model with better diagnosability. A method for quantitative probabilistic sensor fault diagnosability analysis is proposed in this technical note. New definitions of distinguishability are given to measure diagnosability performance among specific faults and fault modes via the principle of weight, which take both noises and distribution laws of probabilistic sensor faults into consideration. Furthermore, some sufficient and necessary conditions for isolating different fault modes are theoretically given. The effectiveness of the proposed method is verified by a numerical example.