Study Of Failure Diagnosis And Expert System For Pressure Vessels

The closed vessel with high-pressure gas or liquid media is generally called pressure vessel which is frequently disabled in accident. It is important that a rapid effective failure analysis is carried out after accidents. As pressure vessel is widely used and efficiency is enhanced, traditional failure analysis is restricted to test, analyze accidents, find out failure reasons and table proposals. It is realistic to make a failure diagnosis expert system for pressure vessels, judge failure reasons rapidly and table proposals. At the beginning, this paper discussed the framework of the failure diagnosis expert system for pressure vessels by studying definition of expert system.Failure diagnosis expert system for pressure vessels is made up of knowledge base, knowledge base management system, inference machine, integrated database, self learning system, explanation machine and human-machine interface etc. Knowledge base and inference machine are absolutely necessary.In the process of making a knowledge base, Fault Tree Analysis is one of effective methods to analyze problems of complicated system, which can reduce the difficulty of obtaining system knowledge in failure diagnosis expert system for pressure vessels. Firstly, this paper discussed failure mechanisms, failure reasons and precautions of explosion, corrosion, abrasion, deformation and leakage etc by studying the rationales of internal and overseas failure analysis for pressure vessels. Secondly, each fault tree of failure analysis was built and the knowledge base of failure analysis for pressure vessels was made by combining production rule with framework in the end.At the same time, this paper discussed the inference machine of failure diagnosis expert system for pressure vessels. The expert system was made with forward reasoning, breadth first search and reliability theory.This failure diagnosis expert system for pressure vessels is easy to run and operate which was designed by using object oriented language. It can distinguish failure patterns, analyze failure reasons and table proposals. This paper gave a typical example and it shows the result is up to the mustard.