Application Of Clustering Algorithm In Rotor Fault Diagnosis

Rotating Machinery acts as an important role in daily life and production, once the fault appears, the related equipment will be affected deathful. Therefore, it is important to monitor the running condition of rotor system, detect faults and analyze its causes. The modern rotating machinery is increasingly sophisticated and complicated, and a lot of complex information is generated, which lead the low accuracy of the current diagnostic tools for fault diagnosis. So, in this paper, clustering analysis will be applied into the rotating machinery fault diagnosis.Firstly, clustering algorithms are studied, three of the classic clustering algorithms for pattern recognition are compared. It is found that K-nearest neighbor clustering algorithm has obvious advantages, then it is put forward that using the K-nearest neighbor clustering algorithm in rotating machinery fault diagnosis. By simulation analysis of K-nearest neighbor clustering algorithm, it is found that the ranging of K value has great influences on the clustering results. The method of using Cross-validation K-nearest neighbor algorithm is proposed to determinate the ranging of K value.Finally, seven faults of common rotor system faults were simulated on the comprehensive experimental platform, through the time domain and frequency domain analysis of collected signal, and get the characteristics of the fault as attribute data, and then the K-nearest neighbor clustering algorithm is used for fault recognition. The results show that the clustering algorithm has certain feasibility in rotor fault diagnosis.