Research On Fault Diagnosis Of Rolling Bearings Based On Generalized Morphological Filtering

Since the industrial revolution,rotating machinery and equipment has been widely used,its safe operation for the production of enterprises is crucial.Rolling bearing,as one of the most easily damaged parts in rotating machinery,has been widely studied in the field of fault diagnosis.The dynamic behavior of the rotor system of rolling bearing is non-linear,the vibration signal is non-linear and non-stationary,the vibration signal is susceptible to ambient noise,these problems increase the difficulty of fault diagnosis of rolling bearing.Therefore,how to extract the characteristics of the vibration signal of rolling bearing and identify the fault type accurately in complex industrial environment is a research direction with practical application value.In order to solve the problem of signal processing and state recognition in rotating machinery fault diagnosis,this paper introduces the generalized morphological filtering method,and studies the problems of signal denoising,feature extraction and fault type identification,and the main research contents include the following parts:(1)A feature extraction method based on adaptive generalized morphological filtering and multi-resolution singular value decomposition(MRSVD)is proposed.In view of the problem that the generalized morphological filtering method is not ideal for noise reduction,the weighted coefficients of generalized morphological filtering are adaptively selected by using the least mean square(LMS)method,which can be used to denoise the vibration signals adaptively.MRSVD analyzes the vibration signal after noise reduction,obtains the fault characteristic information,and analyzes the bearing fault type by Hilbert envelope spectrum.The experimental results show that the method can extract the fault characteristics of the bearing well and verify the feasibility and effectiveness of the method.(2)A feature extraction method based on complementary ensemble empirical mode decomposition(CEEMD)and generalized morphological difference filtering is proposed.In view of the difficulty in separating the characteristics of bearing faults by the generalized morphological filtering method,the generalized morphological difference filter is improved by means of the generalized morphological alternating closed and open method,and the accurate extraction of signal features is realized bycombining with the CEEMD method.The fault signal of inner and outer ring of rolling bearing is analyzed experimentally,and the result shows that the method can isolate the fault characteristic of bearing accurately and has practical value in engineering.(3)In order to improve the identification accuracy of bearing fault types under complex working conditions,a method of bearing fault diagnosis with adaptive generalized morphological filtering and Gath-Geva(GG)clustering is proposed.Compared with FCM clustering and GK clustering methods,the GG clustering method has a higher distinction,the experimental results show that using the GG clustering method to cluster analysis of fault feature samples,the accuracy of bearing fault type identification is improved.