Research On Rolling Element Bearing Fault Prediction And Health Management Based On Deep Learning

Rolling bearing is the most common kind of rotating mechanical transmission device in manufacturing equipment,widely used in wind power,aerospace,mining machinery,textile machinery,automotive parts and other fields.In the actual manufacturing production process,rolling bearings are often in the use of poor environmental conditions and high-speed operating conditions,in the process of operation is very easy to failure.Rolling bearings due to its working characteristics may be caused by a variety of reasons for damage,if the equipment can not be found in time to run abnormal,lightly caused by downtime accidents,bringing losses to enterprise production,serious will also cause safety accidents.Therefore,it is very important to master the operation status of rolling bearings for enterprise safety production.At present,the enterprise has the status quo of relying on manual inspection for the detection of the operation status of equipment containing rolling bearings and manual analysis of the operation data collected by sensors,which requires the operation and maintenance personnel to rely on a large reservoir of professional knowledge in order to analyze the relevant data and judge the operation status of equipment,which does not meet the needs of modern industrial production.In recent years,with the continuous expansion of artificial intelligence application scope,rolling bearing life prediction and fault diagnosis method based on deep learning provides a new idea for rolling bearing equipment health management.Therefore,based on the combination of fault prediction and health management ideas and deep learning,this paper proposes a rolling bearing fault prediction and health management system based on deep learning to realize the control of rolling bearing equipment operation status.Firstly,the rolling bearing failure mechanism and vibration signal analysis method are studied according to the input requirements of the detection model.Secondly,according to the characteristics of rolling bearing degradation trend,the rolling bearing remaining life prediction algorithm based on Tree-LSTM is proposed,and the feature screening layer composed of multiscale convolutional neural network and trend consistency evaluation index is added in the input layer to improve the accuracy of the model for bearing degradation trend,and the percentage error of rolling bearing remaining life predicted by the model reaches 12.39% through the test set verification.It can provide an effective prediction of the remaining life of the bearing.Aiming at the bearing fault classification problem,the rolling bearing fault classification algorithm based on Faster-RCNN proposed,introducing the weighted bidirectional feature pyramid network,and improving the feature extraction performance of the model through the deformable convolution,and the accuracy,precision,recall,and F1 value of the model can reach 95% for ten common rolling bearing faults detection,which can meet the actual equipment operation and maintenance management requirements.Finally,the PYQT software is used to write the terminal upper computer interface of the rolling bearing health management system,and various predictive classification algorithms are integrated to meet the detection requirements under different situations,while improving the automation of rolling bearing equipment operation management and reducing the difficulty of system deployment and operation,which has some practical application value.