Research On Fault Diagnosis Of Rolling Bearing Based On VMD And SVM

As one of the most important parts of modern machinery and equipment,rolling bearing plays an important role in the safety of machine,so it is of great engineering significance to make the necessary fault diagnosis research on rolling bearing.Common intelligent fault diagnosis research steps are: signal acquisition,feature extraction and classification recognition.Among them,feature extraction and classification recognition are the key links of intelligent fault diagnosis.Rolling bearings are affected by the actual working environment,and their vibration signal characteristics are nonlinear and non-steady.Because traditional methods such as time domain analysis,frequency domain analysis and time domain analysis are not effective for complex fault feature extraction,it is very important to choose appropriate fault feature extraction methods.Due to some defects in the commonly used fault diagnosis techniques,rolling bearing fault diagnosis cannot be realized well.Therefore,it is very important to choose the appropriate fault feature extraction method in practical engineering applications.This paper mainly discusses from the following aspects.An appropriate fault feature extraction method can not only extract fault features accurately,but also have a decisive impact on the accuracy of the final fault classification and identification.In view of the above problems,the feature extraction and fault classification of vibration signals of rolling bearings are studied.The findings are as follows:（1）Aiming at the problem that it is difficult to extract fault features due to the non-stationary and nonlinear characteristics of fault feature signals of rolling bearings,a fault feature extraction method based on adaptive variational mode decomposition（VMD）based on improved bat optimization algorithm（IBA）is proposed in this paper.When VMD algorithm decomposes the fault signal,it needs to decompose the number k and penalty factor of eigenmode function α.By using traditional methods of trial and error and single objective optimization to determine the k and pently factor of engenmode decomposition,the optimal combination of parameters cannot be obtained,and the degree of automation is not high,so that its application is strictly limited.Therefore,this paper proposes to adaptively optimize VMD based on IBA to obtain the optimal parameter combination,and then substitute it into VMD algorithm for fault feature extraction.（2）After the fault feature extraction,the obtained fault features are processed.In this paper,support vector machine（SVM）is introduced to construct the fault classification model.Because the kernel function C and penalty factor g of SVM have a key impact on the accuracy of final classification and recognition.Therefore,this paper uses IBA to adaptively optimize the key parameter combination of SVM.The SVM fault classification model optimized by IBA is constructed,and through experimental analysis and comparison with other fault classification and recognition,practice shows that the fault classification recognition model proposed in this paper has obvious accuracy and stability advantages.（3）An adaptive VMD-SVM fault diagnosis method based on IBA is proposed and applied to a laboratory example.