Research On Rolling Bearing Fault Diagnosis Method Based On Improved Empirical Wavelet Transform

Rolling bearings are an important part of rotating machinery,and their health directly determines whether the whole machine can operate normally.Due to the precision of the rolling bearing components and the complicated operating environment,it is prone to failure during work;Once fault occurs,it will cause economic loss at a slight level,and will cause a security accident if it is serious.Therefore,timely and effective condition monitoring and fault diagnosis of rolling bearings are important measures to ensure the safe operation of rotating machinery.When the rolling bearings go wrong,regular vibration will be generated.By detecting the vibration signal,its real-time status and fault type can be judged,thus laying a foundation for the health management of rolling bearings.Therefore,this paper aims at extracting the rolling bearing fault features,Based on empirical wavelet transform to study effective fault diagnosis method of rolling bearing,In order to achieve the effective identification of rolling bearing health status.The main work of the thesis is as follows:(1)For the problem that the empirical wavelet transform based on the parameterless scale-space method is difficult to effectively extract the impact characteristics of the broadband,This paper proposes a parameterless empirical wavelet transform method based on kurtosis optimization.This method uses the sensitivity of the kurtosis index to the impact characteristics,optimize and merge the mode components obtained by decomposing the signal using the scale-space empirical wavelet transform,In order to ensure the integrity of the impact characteristics in the mode component;Then,through the empirical wavelet reconstruction of the mode component with the largest kurtosis index,the effective extraction of rolling bearing fault impact signal features is achieved.The effectiveness of the method was verified by numerical simulation,and it was successfully applied to the fault diagnosis of the outer ring of rolling bearing of SQI test bench.(2)Aiming at the problem of spectrum boundary segmentation based on parameterless scale-space,the wide-band impact feature will be divided into severalfrequency bands,which will cause the mode to break and reduce the accuracy of rolling bearing fault feature extraction.This paper proposes an improved adaptive parameterless empirical wavelet method,Combine spline interpolation fitting with scale-space method to adaptively determine the band boundary of each mode;Then,an empirical wavelet filter is designed based on the determined frequency band boundary to realize the adaptive decomposition of rolling bearing vibration signals.In addition,in view of the problem of the same frequency band noise on the impact feature extraction,this paper further uses a sparse code shrink algorithm to reduce the noise component of the sensitive component containing the impact feature to enhance the periodic pulse feature and improve the accuracy of rolling bearing fault feature extraction.The effectiveness of the method was verified by numerical simulation,and it was successfully applied to the fault diagnosis of rolling bearings on the running parts of electric locomotives.(3)Designed and developed a rolling bearing fault diagnosis system based on LabVIEW platform;Two kinds of rolling bearing fault feature extraction algorithms proposed in this paper are embedded in the system and used for offline data analysis of wind turbine generator front bearings.The effective extraction and fault identification of the fault features of the inner ring of the front bearing of the generator are realized,and the feasibility of the system and the effectiveness of the algorithm are verified.