Research on Low Voltage Switch Mechanical Fault Diagnosis Based on Improved DBSCAN
The traditional low voltage switch fault diagnosis method is based on expert knowledge,has strong subjectivity,re-quires a large amount of fault data as training samples,and has weak generalization ability.This paper proposes an unsuper-vised low voltage switch fault diagnosis method based on density based spatial clustering of applications with noise(DBSCAN).The principal component analysis(PCA)is used to reduce the dimensionality of the low voltage switch coil current signal and obtain low dimensional principal components.The low dimensional principal components are used as inputs for clustering analy-sis in DBSCAN,and the state data are automatically divided into five different types to achieve effective diagnosis of different fault states.Aiming at the difficulty of setting DBSCAN parameters,a fruit fly optimization algorithm(FOA)is proposed to globally optimize it and improve clustering performance.The experimental results based on measured data show that the pro-posed method can achieve better recognition results than 95.6％for all five types of faults,with an average correct recognition rate of 96.5％,indicating higher application prospects.
low voltage switchcircuit breakerfault diagnosisunsupervised clusteringDBSCAN
万方数据
维普
