The Research Of Remote Fault Diagnosis Based On Imbalanced Data Mining

The Remote Fault Diagnosis (RFD) technology integrates the artificial intelligence, network, Data Mining (DM), database and information processing technology. It can solve the complex system diagnosis problem quickly and effectively and improve the diagnosis ability of the whole system utilizing the net transport and knowledge sharing; it is the development tendency of fault diagnosis.Based on the analyses of the requirement of the fault diagnosis system, the RFD System based on DM is proposed. And the system structures and function model of the remote fault diagnosis system are researched; the key technologies of each part of the system were expatiated, the advantage of remote fault diagnosis system is indicated sufficiently. Then the basic framework is proposed for the system that fit for the complex system fault diagnosis.Data Ming is the main method to realize the data amalgamation and knowledge sharing. The Data Mining theory is studied in-depth and the imbalanced data mining algorithm is introduced, using the Support Vector Machine (SVM) with the sampling method together to re-balance data sets, the imbalanced data problem is solved effectively. At the same time, the Extensible Markup Language (XML) is used to manage the database and Web Service technology is used to encapsulate the diagnosis service and the imbalanced data mining module.Finally, based on the Remote Fault Diagnosis System project, a system based on RFD and DM is designed and realized utilizing the ATL and ASP.NET technology, the imbalanced data sets in RFD system is solved. A test example is also been given, results show that the system with imbalanced DM algorithm outperforms the traditional system in terms of both data sharing and diagnosis ability.