© 2016 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works

Distributed fault diagnosis is becoming more and more common in industries, to diagnose faults in any large scale system. There are a lot of disadvantages using centralized fault diagnosis in large-scale systems, since in a centralized implementation all the information has to be collected in one location which is generally not possible or very difficult. Moreover, a centralized system needs a high performance centralized unit which generally in most cases is not available. Due to these difficulties in recent years distributed fault diagnosis techniques have been investigated [10]. In distributed fault diagnosis [1] [2], the global diagnoses for the complete system can be computed from the results in all agents and local diagnose is computed from the results of one agent. In distributed fault diagnosis [3] [8], a global coordination process is not necessary and each subsystem depends on a local diagnoser for local diagnosis tasks and communicating with the remaining local diagnosers until a global diagnosis is produced.

_