Extraction Method Of Weak Multi-fault Acoustic Emission Signals

With the development of modern science and technology,mechanical equipment is towards high speed and heavy load,and it is easy to appear multi-faults with strong noise after long time working.Therefore,it is of great economic significance to identify the weak fault signals and separate them in time for guiding the production and assessing the situation of equipment.Acoustic emission(AE)is a phenomenon of strain energy released rapidly by the elastic wave when the material effects deformation and fracture under external forces or internal forces.As a new nondestructive testing method,AE technology has been widely used in engineering.The AE technique is applied in this paper to monitor the fault signals of the key components of the mechanical equipment.The major investigative contents include:extraction of single source AE signal;separation of weak multi-fault signals for determined condition and separation of multi-fault signals for underdetermined condition.The following achievements had been obtained:(1)Extraction of weak single-source AE signal was studied.An weak signal extraction method of sparse coding shrinkage(SCS)based on independent component analysis(ICA)was proposed.The maximum a posteriori(MAP)method was used to estimate the independent component at first,and the symmetric generalized Gaussian model(GGM)was used to capture the probability density function(PDF)of the independent component.Then the shrinkage function was used for reducing noise.Finally,the estimated independent components were inverse transformed to obtain the denoising signal.By extracting the noisy acoustic emission signal generated by the pencil lead break and tensile crack signal of Q235 steel with different input SNRs,the results are compared with those obtained by the wavelet threshold value denoising method.It shows that the SCS method can extract the tensile crack signal of Q235 steel and acoustic emission signal generated by the pencil lead break with the input SNR is higher than-10 dB,and it is better than wavelet method.The separation results of the method of SCS based on ICA and thewavelet method are similar with the input SNR is higher than 0 dB,the SCS method is superior to wavelet method with the input SNR is lower than 0 dB.(2)Separation of weak multi-source AE signals in determined conditions was studied.An separation method of weak multi-sources fault signals based on wavelet packet analysis(WPA)and ICA was proposed.Wavelet packet technology was used to reduce noise outside the frequency band of the linear mixed signals at first,that is the main frequency of the signals were reserved.Then,the mixed signals were separated by using FastICA algorithm.Finally,the shrinkage function was used for reducing noise within the frequency band.By separating the noisy multi-sources signals consist of acoustic emission signal generated by the pencil lead break and friction signal with different input SNRs,the results show that this method can effectively extract multi-source fault signals with the input SNR is higher than-15 dB.They are higher than those obtained by the method combined WPA and FastICA and only FastICA algorithm.(3)Separation of weak multi-sources AE signals in underdetermined conditions was studied.It includes the estimation of the number of signal sources and the separation of the multi-source signals.An separation method of single channel blind signal separation(SCBSS)method based on EEMD-SVD-ICA was proposed.The ensemble empirical mode decomposition(EEMD)method was used to get multiple intrinsic mode function(IMF)components of the single channel signal at first,then the singular value decomposition(SVD)method was used to estimate the number of source signals of autocorrelation matrix of the IMF components signal.The principal components of the IMF components were as the input signal,and it's number equals the number of source signals.Finally,the multi-fault signals were separated by using Fast ICA algorithm.The method is simple,and it can estimate the number of the signal source and can separate signal.Through separation for single channel signal consisted by two sine signals and two fault signals respectively.The results show that this method can effectively estimate the number of the signal source and separate the signals.And its frequency band is wider,the separation effect is poorer.At the same time,it can be seen that the method can also realize the separation of the crack signal with the frequency band alias.This study provides reference for the early multi-faults identification of the keycomponents of mechanical equipment.It has important significance to improve the safety of the key parts of the equipment and to promote the wide application of AE technology.