Research OnCase-Based Reasoning For Fault Diagnosis

In the modern enterprise, the safety and effectively of the equipment become the base of business development and growth,on the basis of continuous research in the previous already have a set of effective theories and techniques for fault diagnosis, a lot of research direction of the fault diagnosis, there are three categories of analytical model method, the method of signal processing and artificial intelligence diagnostic method, artificial intelligence diagnosis is one of the best. Case-based reasoning for fault diagnosis is a very important direction of research in artificial intelligence diagnosis, by simulating human thinking to resolve failure, from discovery problems and thinking problems, solve problems, remember that the way of thinking of the problem derived basic idea of case diagnosis, making mechanicalfault diagnosis biased intelligent, humane. However, in the case retrieval system, along with the use of the system will continue to accumulate knowledge of the case, if there has not good case study mechanism will lead to a case database sharp growth, which will make the case retrieval efficiency will be greatly reduced.This paper presents a multi-dimensional case-based reasoning model-DRR, The model drop the case of a multi-dimensional space into two-dimensional space point, resulting in the case of two-dimensional clustering, based on this clustering creat R-tree spatial index; while fault case retrieval, rougher search in the R-tree indexes and get the smaller intermediate result sets, and then focus on the intermediate results sets featured case retrieval results. Thissecondary retrieval method not only accelerate the efficiency of case-based reasoning, but also improve the accuracy of the reasoning.In addition, on the basis of the guarantee the accuracy of case-based reasoning, this paper make DRR algorithm to further amend,the multi-dimensional feature weights of the case applied to the case clustering process, in order to obtain a more compact two-dimensional case clustering, further narrowing the size of the intermediate results of case-based reasoning, the DRR algorithm case initial retrieval more efficient, using the concept of secondary weight in case filtering stage to improve the accuracy of case-based reasoning.Finally, the paper applied the fault diagnosis of the case-based reasoning to practical-the fault diagnosis based on a military equipment, research and development of the fault diagnosis system, also designed the case base for this type of equipment failure, and the knowledge representation of case, at last the paper used DRR algorithm and weighting algorithm in this system.