Acquisition Of The Fault Diagnosis Knowlede From Incomplete Information Using Flow Graph

Mechanical devices are widely used in various fields of life and production.Poor running environment and complex conditions will cause the incompletingtesting information of the machinery equipments incomplete. In order to makefull use of the incomplete diagnosing information (IDI), the study of thediagnosis method using the incomplete diagnosing information is needed.In this dissertation, a multi-semantic incomplete diagnosing flow graph isproposed. First, the reasons of IDI are analyzed. Then the definition of IDIsystem is given. By analysing the classical relationship model in the rough settheory and combining the back ground of fault diagnosis and explanation of theincompleting information, the paper proposes a multi-semantic quantifyingtolerance relation model, which is used to analyze the IDI.In order to show the internal relations among incompleting information, thispaper gives the definition of multi-semantic incomplete diagnosing decision flowgraph. The definition of the incomplete attribute value node is also given. At last,this paper gives the calculating formula of the confidence level and coveragelevel, which are taken as the evaluation index of the fault diagnosis rule.Based on the multi-semantic incomplete diagnosis flow graph, the structureof flow graph is reducted through the reduction of attribute layer and attributevalue node. The unnecessary attribute layer and attribute value node are deletedto obtain the most simplified diagnosis decision rule. After so, the mostsimplified flow graph is optimized to obtain the optimal multi-semanticincomplete diagnosis flow graph. Finally, based on the UCI Mechanical analysisdate, the fault diagnosis decision rule obtained from the multi-semanticincomplete fault diagnosis flow graph is tested using the4folds Cross-validationcriteria. The results show that the diagnosis method has a high recognition rateand accuracy of the fault type judgement.