Towards fuzzy partial global fault diagnosis

Abstract

Diagnosis aims at identifying a faulty system based on its behavior observations. It is widely emerged in altogether computer sciences fields, among others: aeronautics, space exploration, nuclear energy, process industries, manufacturing, healthcare, networking, automatism and many other control applications. Diagnosis involves distributed components with an uncertain global view. This paper intends to provide an efficient fuzzy based diagnosis mechanism. Such mechanism enables local hosts’ diagnosis. These local decisions could be merged to provide the global diagnosis. The fuzziness choice is motivated by the fact of incomplete and uncertain system descriptions and observations. Also it is justified by the difficulties of obtaining a complete viewpoint of all system parts where the control is distributed. Our diagnosis mechanism, named FPGD for Fuzzy Partial Global Diagnosis consists of two main steps: Firstly, each remote control host detects and localizes abnormal behaviors which results on a local diagnosis. Each host proceeds by applying a recovery planned actions to maintain system functioning. Furthermore, such local diagnoses should be sent to the global part in order to be merged and analyzed, hence giving a precise and exhaustive global diagnosis. The automatic diagnosis reasoning is a fuzzy system; which based on fuzzy rules, handles incomplete information to deduce system malfunctioning.