Feature Extraction Of Axle Fatigue Crack Acoustic Emission Signal Based On Wavelet Packet Energy Shannon Entropy With A Specific Sub-band

The bogie is one of five major parts of the rail vehicle,and the axle is an important part of the bogie.In the course of the train running,fatigue cracks or even fracture,which will cause serious traffic accidents will appear in the axle due to various reasons.Therefore,it is of great significance to extract the fatigue crack feature of rail vehicle by on-line detection of axle fatigue crack.In this paper,the feature extraction of acoustic emission signal of fatigue crack of axle under various working conditions is mainly carried out.In this paper,a diagnosis plan for feature extraction of acoustic emission signal of fatigue crack of axle is put forward.That is the Shannon entropy of wavelet packet energy in special sub-band.The real time condition of axle is sampled by acoustic emission technology of non-destructive testing.The sliding window is added to the collected acoustic emission signal,and the sliding step size and window width are discussed.The wavelet packet decomposition and EMD decomposition of the signal in the window are carried out respectively.The wavelet base is usually used in fault diagnosis by using Daubechies wavelet as wavelet packet decomposition.According to the characteristics of Daubechies wavelet,Db4 is selected.First,The paper discusses five kinds of entropy and verifies the superiority of the specific sub bands of wavelet packet energy Shannon entropy.The paper Uses wavelet packet energy Shannon entropy(WPESE),wavelet packet energy Tsallis entropy(WPETE)with different non-extensive parameters,wavelet packet energy Shannon entropy with a specific sub-band(WPESE3,2),empirical mode decomposition Shannon entropy(EMDESE),and empirical mode decomposition Tsallis entropy(EMDETE)with different non-extensive parameters in order to extract the feature of normal and fault AE signal of axle.Using evaluation parameter y in order to estimate the of effect of feature extraction of fatigue cracks based on five methods.There fore,We can know optimal method,which is wavelet packet energy Shannon entropy with a specific sub-band;Then,the optimal feature extraction method is used to extract the crack feature of the two conditions(noise,crack signal and hit,crack signal).The results show that wavelet packet energy Shannon entropy with a specific sub-band is the best for the feature extraction of fatigue crack acoustic emission signal.The method can extract the feature of fatigue crack acoustic emission signal in different conditions.