The Study Of The Gear And Rolling Bearings Faults' Signature Seperation Method

The purpose of this paper is to study the methods of extracting the faults'characteristics when gearbox has multifaults by studying the methods of extracting the faults'characterstics when gearbox has single fault.This paper take the gears and rolling bearings as research objects.This paper has simulated some faults of gearbox:the gear and rolling bearings has single fault or multifaults,and analysis the clysical faults and the failure mechaism of the gear and rolling bearings in detail,and had put forward a kind of effective method of rolling bearing's fault diagnosis,Then researched and analysised the diagnosis method of gearbox with multifaults,verified it's validity and practicability .The tested vibration signal of gearbox fault diagnosis generally has cyclostationarity characteristics and modulation phenomenon,so this paper introduced and analysised the theory of cyclostationarity and it's statistics,such as the mean value of a cyclostationary signal, the function of cyclic autocorrelation,the function of spectral correllation density and the degree of the cyclostationary etc.In addition ,introduced the theory of the slice of the cyclostationary spectral.Then combining simulation and experimental data for the research of cyclostationary method of fault feature extraction.Time-frequency analysis is a outlying and high time-frequency resolution method in faults diagnosis of digital signal processing. The paper introduced the basic theory and concepts of time-frequency analysis in detail. Analyzed the linear time-frequency analysis method of short-time fourier transform,and has analyzed in simulation and experimental.In addition,introduced and analyzed the method of bilinear time-frequency analysis,including the most widely used Wigner-ville distribution and the time-frequency distribution of Affine etc.And introduced the rearrangement of spectrums methods of time-frequency distribution in the end.Verified the method of short-time Fourier transform and bilinear time-frequency distribution by simulation and experimental data.Introduced and analyzed methods of restrain the interference and improve signal resolution because cross-term interferences existing in time-frequency analysis.Finally,introduced and analyzed the most important component of the time-frequency analysis:wavelet-analysis. Introduced the basic theories of wavelet analysis,including:the continuous wavelet transform, the discrete wavelet transform, multi-resolution analysis method,introduced the definition of wavelet packet analysis,the method of wavelet packet decomposition and reconstruction, and the wavelet packet band energy method etc. Finally puts forward wavelet packet band energy method with circulation autocorrelation analysis method for analysising gear and rolling bearing's faults.Through the experimental data analysis'results,we obtained that the method for the mechanical failure with modulation phenomenon is always very effective,it is applicable for practical fault diagnosis.Then plant the method of TDAS(time domain averaging across all scales) into the faults diagnosis when gearbox has multifaults,and compared the results of real and complex wavelet functions,then founded that complex wavelet functions is sensitive to signal's frequency, Finaly, diagnosised the rolling bearing's multifaults in gearbox by this method,and verified its'validity and practicability.