Research On Fault Diagnosis Of Rolling Bearing Of Tamping Car Based On Multiwavelet And SVM

The railway,one of the main modes of transport for railroad transportation of goods and people,is considered as the specific manifestation of a country's economic strength.As a major tool for railway operations and maintenance,tamping vehicles play a crucial role.Because the working environment of tamping vehicle is harsh,it is more prone to failure.The rolling bearing is an important working part of the tamping vehicle.If the rolling bearing fails,it will directly affect the accuracy and progress of the entire project.Therefore,how to accurately and quickly find out the failure of tamper rolling bearings has become the focus of future research.In this paper,firstly,the noise signal of the tamped rolling bearing is denoised by the improved multiwavelet threshold denoising method.Moreover,collecting the multiwavelet coefficients in the multiwavelet analysis process,and the relevant statistical value are taken as the fault feature vector.Finally,these vectors of fault feature are classified by the classified models which have been trained.The SVM classification model based on the improved cuckoo algorithm is used to classify fault diagnosis in this paper.According to the poor working environment of the rolling bearing,the collected fault signal contains a lot of noise,which will affect the accuracy of fault diagnosis.To solve this problem,this paper proposes a denoising method to improve the multiwavelet threshold function and takes the multiwavelet coefficient as a fault eigenvector.The eigenvectors are applied to the analysis of fault diagnosis of tamping vehicles' rolling bearings.Due to the multi-wavelet preprocessing method has different pretreatment effects on different vibration signals,in this paper,first of all,for the rolling bearing vibration signal of the tamping vehicle,we select the multiwavelet preprocessing method which is the most suitable of the tamping of the rolling bearing.Then,on the denoising of the soft and hard threshold functions of wavelet analysis,a new multiwavelet denoising method based on improved threshold function is proposed.Through simulation experiments,it's improved thatthe multi-wavelet denoising method for improving the threshold function has good noise reduction effect.And this method is used to denoise the rolling bearing vibration signals.Finally,the relevant calculated values of the multi-wavelet coefficients are used as fault feature vectors.They are applied to the following fault diagnosis.The selection of kernel function parameters and penalty factors in support vector machines is directly related to the accuracy of model classification.Therefore,based on the cuckoo optimization algorithm,this paper proposes a cuckoo algorithm which improve the discovered probability of the nest.This method is used to optimize kernel function parameters and penalty factors.Moreover it selects the optimal parameters to construct a support vector machine classification model.In this paper,the classification model of turret rolling bearing faults based on binary tree-based support vector machines(SVM)has been studied.And the parameters of the SVM will be optimized by the cuckoo algorithm which has been optimezed.Finally,an optimal support vector machine classification model is constructed and applied to fault diagnosis of tamping rolling bearing.The experimental results show that the SVM fault diagnosis model based on the improved cuckoo algorithm can accurately identify the fault category of tamping rolling bearing.