The Research on

Intelligent Fault Diagnostic

Expert System Based on

Fuzzy Neural Network Technology

by Bin Li

Supervised by
Professor Jinpei Wu & Professor Yuanchao  Lei

Dissertation defended in March 1996

Abstract

The research on Intelligent Fault Diagnosis (IFD) is a new tendency of the development of Fault Diagnosis (FD) technology, and today it is being paid more attention to by the experts, engineers and technicians in the fields of Mechanical-Electronic Engineering and Industrial Automation.

In this paper, Intelligent Information Processing (IIP) is introduced to the field of Mechanical-Electronic Device Fault Diagnosis. The three active branches of IIP : Fuzzy Logic (FL), Neural Network (NN) and Expert System (ES) are combined with the traditional FD technology, new classification of IFD is showed clearly, and the realizing technology and methods are also researched in the paper.

At first, a general Fault Diagnostic Model (FDM) is constructed. Based on two-value logic, a general Fault Diagnostic Expert System (FD-ES) is programmed. The program is used to recognize the fault patterns, point out the fixed position of the device fault, and give the fault-removing advices. Secondly, with the reference of the fuzzy properties of FD and the severe defaults of two-value logic model, FL is introduced to design a general Fuzzy Logic Fault Diagnostic (FL-FD) program. This program is used in Fuzzy Pattern Recognition (FPR) of fuzzy logic diagnostic model. According to the fault symptoms which are descripted in fuzzy language, grade of membership of each fault is caculated out. However, the property of fuzzy logic inference is not good, and its abilities of self-studying, self-adaptive and failure tolerance are poor. So taking these causes into account, NN is adopted to act as diagnostic inference mechanism, a Neural Network Fault Diagnostic Expert System (NN-FD-ES) is designed and programmed.

Finally, FL is complemently combined with NN to construct Fuzzy Neural Network (FNN) , and ES is combined with FNN, so these three technology are cross optimized. A Intelligent Fault Diagnostic Expert System (IFD-ES) based on Fuzzy Neural Network (FNN) technology is constructed and realized, and it is the important research content that the paper concentrates on.

Key words: Fault Diagnosis (FD), Intelligent Fault Diagnosis (IFD), Fault Diagnostic Expert System (FD-ES), Fuzzy Logic Fault Diagnosis (FL-FD), Neural Network Fault Diagnosis (NN-FD), Fuzzy Neural Network Fault Diagnostic Expert System (FNN-FD-ES)