Research On Fault Feature Extraction Method Of Rolling Bearing

Modern industrial production equipment is developing towards large,complex,automatic and efficient direction.Equipment investment is huge,and continuous high efficiency production,unplanned shutdown will cause significant loss,so the problem of equipment failure has attracted more and more attention.Fault diagnosis technology has been highly valued at home and abroad,and develops rapidly.As a common and precise rotating machinery component,rolling bearing is prone to failure,and the consequences of developing faults are very serious.It is of great economic significance to detect the state of bearings by machine fault diagnosis technology.In this paper,taking the fault vibration signal of rolling bearing as the research object,aiming at the problem that the fault characteristics of the rolling bearing can not be extracted accurately and quickly under the noise interference,the combination of intrinsic time-scale decomposition(ITD)and the minimum entropy deconvolution(MED)is proposed.The method of feature extraction is also studied,and the application of fast spectral kurtosis graph in the resonance demodulation of rolling bearing fault signal is studied.The specific contents of this paper are as follows:(1)the development of mechanical fault diagnosis and the basic methods of rolling bearing fault diagnosis and the development status of rolling bearing faults at home and abroad are introduced.At the same time,the basic structure of rolling bearing is simply introduced,which lays the foundation for the research of rolling bearing's failure mode and vibration mechanism.The calculation process of theoretical characteristic frequency of rolling bearing is introduced in detail,which provides a theoretical basis for experimental research.It can be known that,although the forms of equipment failure are varied,the rolling bearing failure has the characteristic frequency corresponding to the relatively clear fault characteristics,which provides the basis for the detection work in the project.(2)three typical faults(outer ring fault,inner ring fault,rolling body fault and vibration signals under normal state are measured on the experimental platform by establishing a rotating machinery fault simulation test bench.By analyzing the time domain value of the bearing fault signal and the normal bearing signal,it can be seen that the vibration value of the fault signal of the rolling bearing is greater than that in the normal state of the rolling bearing.Based on this,we can roughly judge whether there is any fault in rolling bearing.But what kind of fault occurs in the rolling bearing,that is,the position of the fault can not be judged only according to the time domain signal of the rolling bearing.Further signal analysis is still needed to determine the exact element of the rolling bearing fault.(3)According to the character of vibration of roller bearing with local fault in strong background noise,an intrinsic time-scale decomposition(ITD)analysis approach combined with minimum entropy decomposition(MED)is proposed.Based on ITD,false components can remove by calculating the noise property of each component and correlation coefficients for the original signal,the signal are reconstructed.The reconfiguration signal were processed by MED to reduce noises.Frequency-spectral analysts for de-noised signal is given to extract effectively the rolling bearing fault feature frequency.The results indicate that this method is able to minimizing background noise and improve the veracity of rolling bearing fault diagnosis.(4)Resonance demodulation is one of the most commonly used methods in rolling bearing fault diagnosis.However,the filter parameters of the resonance demodulation technology need to be set manually,which can only be set according to experience,which brings great contingency and limitation,and needs a lot of experienced workers to execute it.Spectral kurtosis index is very sensitive to the impact characteristics of signals,and is often used in the field of mechanical fault diagnosis.This paper studies and demonstrates the use of fast spectral kurtosis to determine the optimal parameters of the bandpass filter,which makes up for the limitations of artificial setting and reduces the uncertainty of artificial setting.Simulation and rolling bearing fault simulation experiments are carried out to verify the effectiveness of the proposed method.Good results have been achieved.