SDG Fault Diagnosis System Based On Information Granulation Theory And Its Simulation Platform

Zadeh discussed the concept of Granular in 1979, and then Pawlak proposed Rough Set theory in 1982, which could analysis and process inaccurate, inconsistent, incomplete information efficiently. T. Y. Lin proposed Granular Computing(GrC) in 1996, which could be used to eliminate redundant attribute values, extract the minimum attributes set, and achieve useful property from data mining. GrC has become a method of research information and many other problems.When a priori knowledge in system which is not complete or qualitative could not be converted to quantitative information, and a precise quantitative model can not be established, the qualitative fault diagnosis is taken seriously and developed. A branch of the qualitative diagnosis--Signed Directed Graph(SDG) is a root cause analysis, which could reveal fault propagation path, and implement depth analysis and diagnosis of complex failure scenario conveniently. But SDG-based fault diagnosis could generate redundant information, and increase the workload calculation, it needs to be removed. Most superior or a superior reduction can be obtained by Granular Computing reduction algorithm and the heuristic reduction algorithm, which increases attribute according to the attribute important degree.This paper is divided into three major parts. First, SDG-based fault diagnosis and its project application are studied. Second, SDG fault diagnosis Based on GrC and its project application are studied. Third, fault diagnosis simulation platform is built. The following components:(1) In analysis of SDG and SDG-based fault diagnosis theory, an experimental example of a hot nitric acid cooling failure diagnosis system show that SDG model is established according to analysis process system and is simplified. The basic fault diagnosis rules are derived by using the SDG model compatible path, and the fault decision tables are established. To distinguish variables fault bias and retain system information, three states of SDG model nodes is extended to multi-states. In the depth analysis of Granular Computing-based basic theory, information granulation theory is introduced into SDG-based fault diagnosis, the hierarchical fault diagnosis based on information granulation theory and SDG is proposed, using granularity lamination thought based on Granular Computing theory to process SDG model.(2) When fault identification is carried out in SDG fault diagnosis, all nodes and all fault rules may be matched that will reduce the rate of fault diagnosis. All the nodes information is included in fault rule, which will reduce the effective utilization of resource. In analysis of knowledge reduction theory in Granular Computing theory, knowledge reduction algorithm is introduced into SDG-based fault diagnosis, using knowledge reduction algorithm to optimize decision rules of SDG-based fault diagnosis.(3) For industrial production process does not allow fault diagnosis study and testing in the actual process, the fault diagnosis simulation platform based on Graphical User Interface(GUI) is built in Matlab environment, which achieves knowledge reduction algorithm based on Granular Computing, reduces the redundant nodes of fault diagnosis rules, and simplifies the solution of fault diagnosis problems. Simulation platform verifies that the validity and efficiency method, by comparing the reduction fault diagnosis with the original fault diagnosis.