Research On Methods For Fault Diagnosis And Remaining Useful Life Prediction Of Rolling Bearings

As the key component of rotating machinery,rolling bearings are in high load operating conditions for a long time,which can lead to all kinds of failures easily.Once a rolling bearing has failed,it will not only cause economic losses,but may even lead to safety accidents.Therefore,research on the fault diagnosis and remaining useful life prediction of rolling bearings is an inevitable requirement to reduce equipment maintenance costs and ensure reliable operation of equipment.Traditional rolling bearing fault diagnosis methods based on signal analysis require time and frequency domain processing of the signal,relying heavily on human experience,and also suffer from the problem of difficult extraction of fault characteristics.The traditional physical model-based rolling bearing RUL prediction method is more dependent on expert knowledge,and the prediction model established is usually weak in generalization ability.In response to the above problems,this paper uses deep learning related theory for rolling bearings to carry out fault diagnosis and remaining useful life prediction of two aspects of the method research,the main research content is as follows:(1)A fault diagnosis method for rolling bearings based on SDP Images and Mobile Net V2.Aiming at the problems of difficult fault feature extraction and low correct diagnostic rates in traditional rolling bearing fault diagnosis methods,this paper proposes a rolling bearing fault diagnosis method based on SDP images and Mobile Net V2.Firstly,the initial bearing vibration signal is decomposed by Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(ICEEMDAN)and the main feature components are retained to achieve the purpose of denoising.Secondly,the processed bearing vibration signals are transformed into Symmetrized Dot Pattern(SDP)images and input to Mobile Net V2 for extracting fault features and classification identification to diagnose the type of bearing faults.Finally,the effectiveness of the proposed method was verified on the CWRU bearing dataset.Compared with other diagnostic methods,the proposed method has some advantages in terms of correct diagnosis rate and stability.In addition,the noise immunity and generalization ability of the proposed method are verified in the noise addition test and generalization test.(2)A method for predicting the remaining useful life of rolling bearings based on multiscale feature extraction and attention mechanism.Aiming at the problems of difficult identification of degradation stage start points and inadequate extraction of degradation features that occur in rolling bearing remaining useful life prediction methods,this paper proposes a rolling bearing remaining useful life prediction method based on multi-scale feature extraction and attention mechanism.First,the normalized amplitude bearing vibration signal is used as the input,and the quadratic function is used as the label for RUL prediction,avoiding the identification of the beginning of the degradation phase.Secondly,the spatial and temporal features of the bearing vibration signal are extracted using dilated CNN and LSTM respectively,and the attention mechanism is used to assign weights to each degradation feature so as to achieve the RUL prediction of the rolling bearing.Finally,the effectiveness of the proposed method is validated on the PHM 2012 bearing dataset,and the test results show that the proposed method predicts well.In addition,the generalization ability of the proposed method is validated on the IMS bearing data from the University of Cincinnati.