Research Of An Expert System For Automobile Component Failure Analysis And Fracture Image Processing And Identification

As a computer software, Expert System (ES) has some merit such as acute precision, high speed, no tiredness, no bias, etc., which man is inferior to. Along with the coming of information age, the role of ES becomes more and more important. Automobile is a complex mechanical system. It has numerous components, and its failure form is manifold. So it is really significance to research and manufacture an expert system for automobile components failure analysis. As its wide-ranging failure factor, overlapped failure content and fuzzy pattern identification considered, the course of the ES research will be so complicate that it should be gotten on with step by step only. This article make the crack failure of the automobile axle components as the prototype of the ES according to the angle of failure pattern and severity analysis.In the course of building knowledge base, this article analyses field knowledge of the automobile components failure with Fault Tree Analysis (FTA), and represent the field knowledge included in Fault Tree with the combination of frame and production rule on the base of studying the theory and method about knowledge representation.In the course of designing reasoning machine, this article uses forward inference and the tactics of Best-first search. And in connection with much of fuzzy knowledge, it introduces confidence theory to infer the formula of confidence which propagate in AND/OR tree, and illustrates how to calculate the confidence when it propagate in the course of automobile components failure analysis reasoning.Fracture image processing and identification with computer can improve ES' s problem solve efficiency. This article introduces gray image processing technology. Firstly, in order to eliminate noise, optimize image and get precise edge feature, a series of preparations can be done to fracture image such as histogram equalization, binarization, edge enhancement, and so on. Then uses the methods such as Robert, Sobel to detect the edge. Finally, uses the method of Fourier to extract the vein feature of the shell pattern so as to make computer identify the fatigue fracture.