Fault Response Representation And Similarity Measurement Under The Influence Of Variable Working Conditions And Background Noise

Compared with the non-interference environment and steady-state conditions,the rolling bearing vibration signals under variable working conditions and background noise are very complicated.Variable speed,different load,strong noise and other factors will cause the vibration signal to be unstable,variable phase,variable frequency and other unstable conditions,and make its fault characteristics difficult to extract.However,traditional mechanical fault diagnosis has a poor and unstable effect on fault diagnosis under complex conditions and noise.The vibration signal analysis of rolling bearings under variable working conditions and background noise is a hot topic at home and abroad,and has great significance.From the above,my academic dissertation proposes a method based on time series mining to study the homologous analysis of fault response under variable working conditions and background noise.This paper mainly studies the representation of the vibration time series of fault rolling bearings and the similarity measurement method.Because of the high similarity between homologous fault signals,homologous and non-homologous fault time sequences can be used to screen and summarize homologous analysis schemes.The homologous analysis was carried out for experimental data under four conditions;variable speed,different load,outer ring fault position and background noise.The specific research contents are as follows:(1)The time series representation of the fault rolling bearing data of the outer ring,the rolling element and the inner ring is used to studied the characteristics of different time series representation methods.The advantages and disadvantages of Second Fast Fourier Transform,Phase Space Reconstruction,Cepstrum and Piecewise Linear Representation are summarized.This paper studies the characteristics of different similarity measures for the similarity measure of simple time series.The advantages and disadvantages of Euclidean distance,JS divergence,Symbolized distance,Cosine distance and Dynamic time warping are discussed.(2)The homologous analysis scheme was used to analyze the experimental data of the rolling bearing with different diameters of the outer ring of the Western Reserve University,and then verify and summarize the effectiveness and characteristics of different schemes.The best solution is Second FFT combined with JS divergence and Second FFT combined with DTW.Finally,the best solution was used to verify the reliability of the experimental data of Case Western Reserve University outer ring,ball and inner ring fault rolling bearing.(3)Aiming at the diff-iculty of analyzing mechanical vibration signals under background noise under variable working conditions,using experimental data,Case Western Reserve University data and white noise signals to investigate the effects of variable speed,different loads,different signal-to-noise ratios and different outer ring fault locations on the homology analysis of fault response.The results show that the scheme(secondary FFT combined with JS divergence)is still stable in the case of high rotational speed,and has advantages in comparison.In experiments with different loads,the resolution of the second FFT combined with DTW is better and more effective.In the experimental analysis with different signal-to-noise ratio,the stronger the noise,the more severe the scheme(second FFT combined with JS divergence)is disturbed by noise,and the effect of the scheme(second FFT combined with DTW)is more stable.In this paper,the experimental data of different locations in the outer circle were analyzed.It was found that the homologous similarity was the highest at 12 o’clock and the lowest at 6 o’clock.