Identification Of Wavelet Energy And Shannon Entropy For Feature Extraction In Axle Fatigue Crack Acoustic Emission Signal

The axle of railway vehicle is a important part of the bogie.In railway vehicles long running process,the axle applied external load by the body can appear fatigue crack,and the crack will cause the energy accumulated in the material to be released in the form of elastic waves,producing acoustic emission phenomenon.Thus,in this paper,acoustic emission testing technology of non-destructive testing technology is used to detect the damage of the axle of railway vehicle.Wavelet energy and Shannon entropy are used to extract the feature of AE signal of fatigue crack of axle and describe the extension process of fatigue crack of axle because their distribution of the axle fatigue crack AE signal are different to noise signal in time-frequency domain.The axle fatigue crack will undergo different stages from initiation to extension to axle fracture.This paper will be collected by the fatigue experiment all process of the axle fatigue crack AE signal segmenting processing according to the waveform characteristics of AE signals preliminary divided into four groups.Morlet wavelet is suitable to extract the impulse components of mechanical fault signals and thus its continuous wavelet transform(CWT)has been successfully used in the field of fault diagnosis.Thus,in this paper,each group AE signals is processed with CWT and wavelet coefficients of each group AE signal are obtained and the Morlet wavelet is choosed as wavelet basis function of CWT.The paper choose five characteristic parameters,average,peak-peak value,standard deviation,skewness,and kurtosis factor,extract the feature of wavelet coefficients of different resolution scale to constitute the feature vectors.the paper proposed a method,which is the optimal range of wavelet scales selected based on the maximum energy to Shannon entropy ratio criteria and consequently feature vectors are reduced.In addition,wavelet energy and Shannon entropy of the wavelet coefficients are used as two new features along other statistical parameters as input of the classifier.Finally,a multilayer perceptron neural network is used to classify the input features.The results based on verification analysis show that continuous wavelet transform(CWT)based on Morlet wavelet energy and Shannon entropy on the optimal scale distribution not only can extract the feature of the axle fatigue crack AE signal from hit and noise signal,but also can show the different stage of axle fatigue crack.the choice of the optimal scale range reduces the feature input.The addition of new features improves the performance of the classifier,and can identify the axle fatigue crack.