Research of the Fault Testing and Diagnosis Methods Based on Data Mining
|School||Xi'an University of Electronic Science and Technology|
|Course||Circuits and Systems|
|Keywords||Fault Diagnosis Data Mining Virtual instrument Information Fusion System implementation|
This thesis in data mining, virtual instruments and information fusion theory based on depth study of complex systems fault diagnosis method and its application . The study of large complex systems efficiently , safe, stable and reliable operation , improve of equipment production efficiency and management level , the maximum extent of recovery faulty equipment and restore the economic losses caused by the failure to meet the sound and fast economic development has is of great practical significance . The complete fault data collection experiment based on virtual instrument designed in this paper provide sufficient data for a follow-up study . For complex systems , focusing on rough sets algorithm of attribute reduction and feature extraction , the application of the decision tree algorithm in mining malfunction rules and clustering algorithm incremental data mining application of the new fault rules . By improving rough set attribute reduction algorithm to solve the problem of attribute reduction in less efficient ; directly the fault database for efficient data queries and processing , greatly improving the the ID3 algorithm efficiency and can achieve through the use of embedded SQL ; by the use of the the ART2 algorithm with K-means algorithm combined method effectively inhibited ART2 drift of the cluster center . Information fusion fault diagnosis , diagnostic data fusion , making the diagnosis more comprehensive data ; integration of diagnostic methods , makes diagnosis more rapid , accurate and reliable results . Finally, this three algorithms as the core , the hydraulic system fault diagnosis application background test of a fault diagnosis system design and implementation . Comprehensive use of various functional modules to achieve fault data collection , storage , pretreatment , rules mining, fault diagnosis and report generation print a series of functions . Equipment failure diagnosis experiment , it is proved that the system can accurately carry out fault diagnosis , while rapid self-learning ability to discover new rules , to better meet the needs of the actual fault diagnosis .