Rolling Bearing Fault Diagnosis Based On Case-based Reasoning

Based on the research on fault diagnosis technology in common use, using artificial intelligence, fault diagnosis of theories and methods, the paper researched and developed a rolling bearing fault diagnosis expert system prototype, in order to achieve the intelligent diagnos for rolling bearing.In-depth study of the commonly used intelligent fault diagnosis technology and case-based reasoning method , a design and prototype system is proposed based on the rolling fault diagnosis expert system. The expert system analysis a lot of fault maintenance log and characteristics of the rolling bearing fault diagnosis , explore the key structural expert system technology, including case-based reasoning of the case retrieval of knowledge engineering and technology.Detected the rolling bearing vibration signals by acceleration sensors ,used wavelet packet analysis method to analysis of vibration signals, a fault energy eigenvector is extracted. This feature vector as the case ,built the case library.After comparing the VA-file method , VA+-file method and multi-resolution VA-file , decided to adopt the multi-resolution VA-file method based on the use of the wavelet transform to tectonic the indexing documents of the rolling bearing diagnosis case library. That is ,toWavelet transformed first-dimensional vector as the principal component,then to sort the principal component in order to establish index.This method can reduce the the number of visit to approximate vector, increase speed and efficiency of case search.Finally, based on the needs of Rolling fault diagnosis system, the overall structure and software architecture is designed, and some of functional module and future work is introduced in this paper.