Fault Diagnosis Of Rolling Element Bearings Based On Neighborhood Rough Set And Random Forest Classifier

Rolling bearing is one of the most used parts in the rotating machinery equipment of Qingdao special steel.When the rolling bearing failure,if it is not timely replacement,is bound to cause mechanical equipment downtime,affecting the actual production schedule.However,how to judge the fault state of rolling bearing effectively is a key problem to be solved.In order to enrich the Qingdao special steel power equipment department of rolling bearing fault diagnosis scheme for fault diagnosis of rolling bearing,thinking to broaden the current monitoring system,based on the rolling bearing in high speed wire rod mill of air-cooled line fan equipment as the research object,first of all,from the time domain,frequency domain and time-frequency domain to extract fault feature composed of 31 original fault the feature set;secondly,using neighborhood rough sets of fault feature from original fault features were identified in 12 sensitive fault characteristic parameters,reduce the subsequent recognition burden;then,using the random forest classifier combination as a classifier to distinguish the fault state of rolling bearing four common fault state,at the same time,according to the decision tree random forest classifier the number and the split attribute number the two parameter setting problem,A parameter optimization algorithm based on genetic algorithm is proposed.Finally,from the correct diagnosis rate,considering the training speed of two angles,and learning vector quantization network fault diagnosis as state classifier results were analyzed to verify the validity based on neighborhood rough sets and fault diagnosis method of rolling bearing random forest classifier.