Research On Intelligent Fault Diagnosis Method Of Rolling Bearings Based On Deep Learning

As a key supporting component in rotating machinery,the health status of rolling bearings directly affects the performance of mechanical equipment.In order to effectively avoid unnecessary loss of life and property,rolling bearing is implemented timely,accurate and efficient fault monitoring and diagnosis.At present,with the increase of equipment refinement and the complexity of working conditions,fault diagnosis has been put forward new challenges.With the rapid development of state monitoring technology,the sampling points,sampling periods and the amount of data collected in the monitoring system are increased,and the processing methods of fault data and fault diagnosis algorithms are put forward higher requirements.Rolling bearing is taken as the research object in this paper.Convolutional neural network,Kurtogram and deep separable convolutional neural network,convolutional neural network and long short-term memory network are used to carry out intelligent fault diagnosis based on deep learning.The main content included the following three aspects:(1)Aiming at the problem of low recognition accuracy of traditional intelligent diagnosis methods,a rolling bearing fault diagnosis method based on convolutional neural network is proposed.Firstly,the vibration signal of rolling bearing is collected after the gearbox reliability data acquisition system is built,and the data is normalized and enhanced.Then build a convolutional neural network model,filter the local area of the input data through the convolutional layer,further reduce the dimensionality of the input data through the pooling layer,fuse these abstract features through the fully connected layer,and classify by the Softmax activation function.Through experiments to verify the performance of the convolutional neural network.The results show that the convolutional neural network has a high diagnostic accuracy rate in the field of bearing fault diagnosis.(2)Aiming at the problem of small sample size in actual diagnosis problems,which is not enough to stabilize training.The sensitivity of Kurtogram to noise-containing signals is comprehensively considered,and the great advantage of deep-separable convolutional neural network in image recognition,the K-DSCN model is proposed for fault diagnosis of rolling bearing.Bearing fault diagnosis.The Kurtogram technology is used to convert the time-domain vibration signal into a two-dimensional graph and the Kurtogram graph is used as the input of the deep separable convolutional neural network for intelligent fault diagnosis.The proposed method is compared with a variety of methods.The results show that Kurtogram can effectively extract the potential information of bearing vibration signals with different fault types;compared with the traditional convolutional neural network,the deep separable convolutional neural network can greatly reduce the number of network parameters,and can more effectively extract features from the two-dimensional feature graph;compared with other models,K-DSCN model has high classification performance and better generalization performance.(3)In order to improve the efficiency of fault diagnosis and enhance the stability of the model,a fault diagnosis method of rolling bearing based on CNN-LSTM is proposed.Through the convolutional layer and pooling layer of the one-dimensional convolutional neural network,local spatial feature extraction and dimensionality reduction are performed,and the long short-term memory network is used to extract temporal features.In order to improve the sensitivity of the network to data,multiple strategy optimization models such as Dropout,Mini-batch,L2 norm,and BN are used.The experimental results show that:compared with other models,the CNN-LSTM model has a very strong generalization ability.