Research And Implementation Of 6205 Type Bearing Fault Diagnosis Algorithm

With the development of automation technology in the industrial manufacturing industry,from the perspective of safety,machinery and equipment often need to remain in a stable working state.Rolling bearing widely,whether in common mechanical equipment or efficient and precision mechanical equipment is an indispensable part,any bearing damage will be in production and life bring serious problems and hidden trouble,so in the process of production and life of rolling bearing fault diagnosis,determine its working condition is extremely important.Since the results of the rolling bearing signal will have a direct impact on the diagnostic results,the fault diagnosis method of the type of rolling bearing is studied.The main contents are as follows:(1)Aiming at the problem of affecting fault diagnosis noise in the vibration signal,an improved wavelet packet noise reduction reconstruction algorithm is studied.Firstly,the vibration signal is decomposed by wavelet packets,and the sub-frequency band is screened and reduced based on the law of margin of the correlation number,and the reconstruction of the sub-frequency band is completed to obtain the reconstructed signal after the noise reduction process,and the energy percentage between the reconstructed signal and the original signal is used as a measure of the quality of the noise reduction effect,and the ratio is 92.675%.It is validated using the bearing dataset to extract valid feature information.At the same time,anti-noise experiments are carried out to prove the effectiveness of the proposed method in noisy environments.(2)Aiming at the problem of low diagnostic accuracy caused by the single input feature of the traditional diagnostic model,a double convolutional neural network model with feature fusion is studied.The network model contains two inputs,one-dimensional vibration signal and two-dimensional time-frequency pattern as inputs,in order to suppress non-important features in the time-frequency pattern,add an attention mechanism module to the network,and finally fuse all the features in the full connection layer.In this paper,the model is verified by using the relevant public data set,and the verification result reaches 98.76%,which proves the effectiveness of the model.(3)Starting from practical applications such as industry,the above fault diagnosis algorithm is designed as a system.The individual modules are integrated in this system to implement their diagnostic functions.Including the time domain,frequency domain and time frequency domain analysis of the signal,while obtaining the signal time-frequency map,its threshold noise reduction processing and other operations,convenient for users to perform multi-angle analysis of the signal,and finally get diagnostic results.The system is easy to operate and has certain practical value in industrial applications.