Collection And Verification Of Rolling Bearing Fault Data Set

Rolling bearing is one of the most widely used, critical and easy to be damaged parts in the mechanical equipment, and its running status is directly related to the whole machine or the whole system performance. According to the statistics, there are about30%mechanical failures caused by rolling bearing in rotating machinery. Therefore, the rotating machinery fault diagnosis of rolling bearing has been the focus and difficulty of the research for a long time. The more mature application is just for single fault under uniform motion in rolling bearing fault feature extraction at the present stage, but start and stop frequently in the variable speed state is not in the minority in the actual operation. And also some bearing damage to be a certain degree will be replaced in the practical engineering equipment, there may exist the phenomenon of the coexistence of multiple bearing faults during the period. At the same time, multiple fault detection is still a great challenge in monitoring and diagnosis of rotating machinery. Multiple fault diagnosis mainly concentrated in that there is a fault in different parts of the bearing at the same time, rotor system touch friction and crack, shaft misalignment and rotor imbalance of compound fault. According to the report, work for rolling bearing fault detection at the same time is very limited so far. Therefore, a mature multiple fault data set becomes especially important.The rolling bearing common failure form, rolling bearing kinematics and the local fault bearing experimental design were firstly introduced in this paper. Rolling bearing fault vibration signal under the condition of constant and variable speed were collected in the ST-5000A Beijing Jiaotong University multifunctional rotor experimental platform and Shanghai University BVT series of rolling bearing vibration measuring instrument,which ensure that the theory research work is based on experimental verification. At the same time, we collect a relatively complete set of experimental data set of fault rolling bearing.On the basis of the above experimental data.Based on the spectral kurtosis of resonance demodulation method and EMD empirical mode decomposition of the resonance demodulation method to process and analyze uniform multi fault vibration signals, verified the quality of the collected rolling bearing fault uniform data in this paper.Based on rolling bearing of instantaneous fault characteristic frequency trend line structure and extraction, which is based on instantaneous frequency estimation method for extracting speed verified in the absence of speed information.This method verified the quality of the collected rolling bearing single and multi fault variable speed data which contain only the faulty bearing vibration information in this paper.Finally,based on spectral kurtosis of the rolling bearing fault envelope order tracking analysis method is used to verify the quality of the collected multiple faults variable speed vibration signal data which have speed information, and improve and optimize the processing flow of the variable speed multi fault method to make it become more effective.