Research Of Electromagnetic Acoustic Emission Signal Denoising Based On Wavelet Theory

The electromagnetic acoustic emission nondestructive testing technology is newlyemerging in recent years. At present it is still in the processing of experiment. Noiseinterference cannot be avoided during the collection of electromagnetic acoustic signals. Soin order to improve the accuracy for electromagnetic acoustic emission nondestructivedetection, denoising processing becomes an important research topic. Electromagneticacoustic emission signals are real transient, non-stationary and time-varying. To deal with it,traditional Fourier transform cannot have good effection on it. Wavelet transform has beenwidely used in signals processing due to its good characteristics in time-frequency analysis.But the wavelet transform focuses on low-frequency analysis, cannot deeply analyze highfrequency processing.The wavelet packet transform has a good improvement on high frequency analysis.However, when different wavelet packet deals with the same signal, the noise reduction hasobvious different effect. Lots of experiments need to be done to deal with the signalanalysis and it is complex. Wavelet packet, which has a tight frame, the higher vanishingmoment and the shorter supporting length, is suitable for processing the electromagneticacoustic emission signal. So the wavelet packet library for db1~db20, sym1~sym20isselected here. This paper presents adaptive wavelet packet. Different energy priorities havebeen adopted. Two Adaptive wavelet packet methods are carried out twice. Combined withthe wavelet packet threshold method, an improved threshold function is proposed by usingthe signal itself as the asymptote and an improved estimation of threshold is proposed byreplacing uniform threshold estimation of the median with the mean of wavelet packetcoefficient absolute value. The final thresh estimation is obtain by combining theinformation of each frequency band.The experimental signals are collected by using4channels of PCI-2system and thewhole program runs by MATLAB. The experimental results show that the new denosingmethod is effective for the electromagnetic acoustic emission signal and can provide a toolto study. the electromagnetic acoustic emission nondestructive testing technology.