SSL4-1800 Motor Bearing Fault Diagnosis And Residual Life Prediction For Stamping Equipment

Stamping motor bearing is an essential component in servo press,and its running state directly affects the running stability of servo press.In the process of system operation,poor lubrication and excessively high temperature may lead to bearing failure or degradation.Once the bearing fails,the stability of equipment will decrease or even stop.Timely understand the bearing running condition,can help the motor maintenance staff maintenance and repair in a timely manner,to prevent the stamping workshop production loss has important research significance.Motor bearing vibration signal based on the stamping equipment,stamping equipment motor bearing fault diagnosis and remaining life prediction,the main work of this paper is as follows:(1)Aiming at the problem that vibration signals of motor bearings contain too much noise,a EAVGH-SWD bearing fault diagnosis method was established in this paper.Firstly,the optimal scale of structural elements was selected by using the characteristic energy factor(FEF).At the optimal scale,the signals were enhanced and denoised by the EAVGH enhancement operator.Will feature the enhanced signal decomposition(SWD)decomposition,selected by kurtosis value most OC component contains information in the original signal,on the envelope analysis of the servo press motor bearing fault diagnosis.Through stamping equipment motor bearing outer ring fault simulation signal analysis and experimental verification,the results prove that the method is effective to extract the stamping equipment motor bearing fault feature of vibration signal.(2)For stamping equipment motor bearing vibration signals is difficult for the early fault feature extraction problem,is constructed based on CNN-TCN-Attention bearing the residual life prediction model of the network.Through EAVGH hat operator characteristics of vibration signal is enhanced,will signal after filtering by convolution neural network(CNN)mining in the deep features of the signal,faint deep fault features are extracted from the signal,And by the time,convolution network(TCN)combined with Attention mechanism(Attention)the key features of the choice of the signal effectively,improve the prediction precision,through contrast experiment and different model,verified the method has high prediction accuracy.(3)Finally,the fault diagnosis model and residual life prediction model established in this paper are applied to the motor bearing signals collected in practice,which verifies the good engineering application of the model established in this paper.