Identification of Metal Crack Signal of Deep Drawing Based on Wavelet Packet and AR Spectrum Analysis

Abstract

According to the characteristics of AE signals of the metal early micro crack in the process of deep drawing, the parameters to identify the crack signal are selected. The wavelet packet are adopted to resolve the AE signals with complex background noises and feeble crack characteristic, as well as the time series analysis method is applied to established the AR model of the resolved signal and to extract the energy value of the AR spectrum. The characteristic parameters are depended on the ratio of the energy value of the resolved signal bands and the total energy value of the crack AE signal. Finally, the method of fuzzy comprehensive evaluation is used to detect the crack signal by comparing the five models of evaluation results. Experimental results show that the application of the above methods have an unparalleled advantage on identifying the early micro crack signals with short-term impact character.