Research On Extraction Technology For Fault Feature Of Rolling Bearing

As a kind of commonly parts used parts in the mechanical equipment,the rolling bearing will directly affect the normal operation of the mechanical equipment,so that it is very important to monitor and diagnose the bearing.As the key link of fault diagnosis feature extraction is the main content of this paper,in this paper,a new method of fault feature extraction based on approximate entropy and square demodulation analysis,sample entropy and fractional Fourier transform,fuzzy entropy and envelope analysis is studied.In addition,also put forward the entropy characteristic parameter method,and combined with the support vector machine to identify the fault pattern of rolling.Specific research contents are as follows:(1)A new method of fault feature extraction based on approximate entropy and square demodulation is proposed.Firstly,the concept of approximate entropy is introduced,and the feasibility of using approximate entropy to characterize bearing fault characteristics is verified by analyzing the vibration signals of different fault modes.At the same time,the simulation results show that the square demodulation technique can effectively separate the modulated wave from the high frequency modulation signal.Finally,a new method of feature extraction based on approximate entropy and square demodulation is studied,and the different fault modes of rolling bearing are distinguished according to the approximate entropy of the square demodulation;In order to verify the adaptability of the method,the vibration data of different load,different damage degree and different sample length are studied.(2)A new method for fault feature extraction of rolling bearing based on sample entropy and fractional Fourier transform is proposed.Firstly,the concept of sample entropy is introduced,and compared with the approximate entropy in the aspect of noise resistance and requirements for sample length.At the same time,the definition of fractional Fourier transform and its good time-frequency analysis are discussed in detail,then taking the vibration signal of rolling bearing as an example,the energy concentration of the fractional Fourier transform and the classical Fourier transformare compared.Finally,a new feature extraction method based on sample entropy and fractional Fourier transform is studied,and the different fault modes of rolling bearing are distinguished according to the sample entropy of fractional Fourier transform;At the same time,it is proved that the sample entropy based on the optimal fractional Fourier transform is insensitive to the noise interference and the sample length.(3)A new method for fault feature extraction of rolling bearing based on fuzzy entropy and envelope analysis is proposed,and on this basis,recognition of multi entropy method is proposed.Firstly,the concept of fuzzy entropy is introduced,the simulation results show the effectiveness of the proposed method in the detection of mutation and state recognition,and compared the performance with the sample entropy.At the same time,the simulation and experimental results show that the envelope demodulation technique can effectively separate the modulated wave from the high frequency modulation signal.Then,a new feature extraction method based on fuzzy entropy and envelope demodulation is studied,and the different fault modes of rolling bearing are distinguished according to the fuzzy entropy of envelope demodulation;Finally,combine with the approximate entropy of the square demodulation,the sample entropy of the fractional Fourier transform and the fuzzy entropy of the Hilbert envelope,the recognition of multi entropy method is proposed,and combine with the support vector machine to identify the fault mode.