Research on Condition Monitoring of Bearing Based on Factor Analysis and Logistic Regression

Abstract

Along with the introduction of the concept of industrialization 4.0, we are taking another step towards the times of big data. The informationized, datamation and intelligent development of industry pushes the development of monitoring on rail traffic health, which makes accurate and effective monitoring on trains status and prevention of severe accidents become an inexorable trend. The research object of the paper is rolling bearing which is one of the key components of pulling motor. The research process it that collect the vibration signals of bearing through acceleration sensor, pretreat the raw data, then extract the feature indexes, carry out factor analysis on feature indexes and research into the correlation among feature indexes, surmount the subjectivity for index selection, then realize dimensionality reduction of big data, establish the bearing status deterioration assessment model with logic regression and get the health degree through calculation of real-time feature indexes, then monitor the operation state of bearing according the health degree , and find problem timely, thus reduce the accident incidence and maintenance cost and raise economic benefit.