Research On Early Fault Diagnosis Of Rolling Bearings Based On Deep Learning

Rolling bearings are one of the important components in mechanical equipment,and their health has a significant impact on the stable operation of the entire equipment.Once the rolling bearing malfunctions,the equipment will stop operating and cause economic losses,and the equipment will be damaged and cause safety accidents.Therefore,condition monitoring and fault diagnosis of rolling bearings are the key to ensuring the safe operation of equipment.With the development of technology,the rolling bearing condition monitoring signal acquisition system is becoming more and more mature.How to extract fault information from the monitoring signal and give an accurate judgment has become an urgent problem to be solved.Deep learning is different from traditional diagnosis methods.It can automatically learn the inherent characteristics from massive data to obtain a good feature expression,which brings new ideas for the diagnosis of rolling bearing faults.This paper studies the fault diagnosis method of rolling bearing based on deep learning.The main research contents are as follows:(1)Analyze the research background and research significance of rolling bearing fault diagnosis,discuss the characteristics of rolling bearing fault diagnosis method based on deep learning,and analyze the existing bearing fault diagnosis method based on deep neural network.(2)Construct the framework of fault diagnosis model.In view of the difficulty of the traditional rolling bearing fault diagnosis method in the feature extraction of massive data,a rolling bearing fault diagnosis model based on restricted Boltzmann machine(RBM)and autoencoder(AE)is proposed,and the batch standardization strategy(BN)is introduced.)Improve the convergence speed and accuracy of the model.(3)Add noise to the data set and perform anti-noise training on the model to improve the classification ability of the model in a complex environment.The experimental results show that the method can accurately locate the fault location of the bearing,at the same time accurately identify the fault degree,and has a certain degree of noise resistance.Its feature extraction and recognition capabilities are stronger than Back Propagation Neural Network(BPNN),Convolutional Neural Network(CNN),Deep Belief Network(DBN)and Stacked Auto-Encoding Network(SAE).(4)Aiming at the problem of the lack of data samples of the equipment to be diagnosed,the algorithm model is combined with the transfer learning idea,and the bearing failure public data set is used as the auxiliary sample,and the data collected by the experiment is used as the target data sample for research.Experiments show that the classification ability of the model combined with the idea of transfer learning is better than the original model.(5)Using tools such as Pycharm,wx Form Builder and My SQL database to design and implement a rolling bearing fault diagnosis system.It has realized bearing fault diagnosis,data management and other functions,which can assist the staff to identify the fault type of the bearing.