Research On Feature Extraction Method Of Rolling Bearing Vibration Signal

Nowadays,with the rapid development of science and technology,rotary machinery and equipment are constantly developing in the direction of high speed,heavy load and precision.Rolling bearings are an important part of rotating machinery.The operation of rolling bearings directly affects the performance and personal safety of the equipment,the entire unit,and the entire production line.Therefore,research on fault diagnosis of rolling bearings is of great significance for maintaining human-machine safety and improving production efficiency.Feature extraction is a key step in fault diagnosis.How to extract signal features that can effectively reflect the rolling bearing operating conditions has theoretical significance and engineering value.The rolling bearing fault diagnosis and feature extraction are taken as the research objectives,and the feature extraction method of the rolling bearing vibration signal is taken as the research content,and the bearing fault feature information extraction is analyzed in this thesis.Aiming at the problem that the fault feature information of rolling bearings is often submerged by strong background noise,a feature extraction method based on peak index,wavelet decomposition and Hilbert envelope spectrum analysis is proposed.Firstly,the peak index is used to characterize the transient fault size of the rolling bearing vibration signal.Then,the vibration signal is decomposed and reconstructed by wavelet transform.Finally,the Hilbert envelope spectrum analysis method is used to transform the detail signal into spectrum.The results show that early failure characteristic frequency of the rolling bearing under the strong noise background can be extracted effectively using this method.Combining the time-frequency distribution characteristics of vibration signal and information entropy theory,a feature extraction method combining EMD algorithm and sample entropy is proposed,solving the problem of redundant attributes in rolling bearing fault feature information set.Firstly,the original signal is decomposed by empirical mode,and it is decomposed into several stable intrinsic modal functions.Then,the mutual relationship,kurtosis,and variance of each IMF component are calculated,and several IMFs containing major fault information are selected and analyzed.Finally,the sample entropy value of each IMF component is calculated,and the bearing vibration signal feature is extracted using the sample entropy value.The results show that the signal characteristics that can reflect therolling bearing operating conditions can be extracted effectively using the EMD sample entropy method.For the problem that early fault characteristics of the fault bearing vibration signal is difficult to extract.The feature extraction method of rolling bearing combined with wavelet packet transform and hierarchical entropy is proposed in this thesis.First,three-layer wavelet packet decomposition is performed on the vibration signal.Then,the hierarchical entropy of different frequency bands is calculated.Finally,the characteristics of the rolling bearing vibration signal are extracted by the wavelet packet level entropy value.Compared with the traditional fast Fourier transform method,the wavelet packet hierarchical entropy method can effectively improve the accuracy of feature extraction and describe the characteristics of the rolling bearing vibration signal more precisely and completely.the accuracy of feature extraction can be improved effectively and the characteristics of the rolling bearing vibration signal can be described more precisely and completely using the wavelet packet hierarchical entropy method.