Rolling Bearing Fault Diagnosis Based On Vibration Signal Analysis And Neural Network

Rolling bearing is a vital part of machinery and equipment,and its working condition will directly affect the operation of the whole equipment.In this paper,the rolling bearing is taken as the object.Starting from the three parts of fault signal acquisition,fault feature extraction,signal noise reduction and fault type recognition,the fault feature extraction and recognition are mainly discussed.A rolling bearing fault diagnosis method based on vibration signal analysis and neural network is proposed.The main research contents are as follows :A rolling bearing fault feature extraction technology based on ALIF-SVD is proposed.Firstly,the adaptive local iterative filtering(ALIF)algorithm is used to process the fault signal to obtain several intrinsic mode functions(IMFs).After calculating the sample entropy,the threshold is set for signal reconstruction.Then singular value decomposition(SVD)is carried out to draw the difference spectrum curve.Finally,the secondary reconstruction is performed according to the mutation position in the differential spectrum to further complete the noise reduction.In the fault feature extraction model based on ALIF-SVD,the ALIF algorithm has an adaptive filter,which can more accurately capture the local instantaneous frequency of the signal;the SVD algorithm has a significant noise reduction effect and can truly and effectively reflect the characteristics of the measured signal.This model not only greatly improves the modal aliasing problem existing in other methods such as Iterative Filtering(IF),but also solves the common problem of redundancy of a large number of noise signals.Through simulation experiments,the test accuracy,signal-tonoise ratio and time of the model and the basic algorithm model are compared.It is proved that the model can extract useful signal features more accurately and quickly,which reflects the effectiveness of the model.A rolling bearing fault diagnosis technology based on FA algorithm optimized BP neural network is proposed.In order to solve the problem that BP neural network is easy to fall into local extremum,a method of optimizing BP neural network based on Firefly Algorithm(FA)is proposed,and a FA-BP model is established for fault diagnosis and classification of rolling bearings.The vibration signal of rolling bearing after noise reduction is processed,and the data set is randomly divided into training set and test set.The model is used to diagnose the fault of rolling bearing.In the rolling bearing fault diagnosis model based on BP neural network optimized by FA algorithm,FA algorithm has the characteristics of global optimization,which can effectively solve the problems of local extremum and error convergence of BP neural network.Through simulation experiments,the FA-BP model and the commonly used classification model are analyzed,and the average classification accuracy and the average time required for the model are compared.The experimental results show that the proposed FA-BP model method has better classification characteristics.The model has higher recognition accuracy,faster operation speed,and more accurate fault detection of rolling bearings.