基于小波包变换的金属材料超声防伪识别

摘要

Three kinds of metal materials with different composition and three kinds of metal materials with similar composition are taken for samples in this work, and the 10 MHz ultrasonic scattering signal in the samples of the materials was extracted.The anti-counterfeiting features in the scattering signal can be got by the wavelet packet transform.The genetic algorithm was used to optimize BP neural network as classifier.Results show that the metal materials with different composition and the metal materials with similar composition can be identified successfully.Compared with the similar composition of metal materials without heat treatment, the heat treated ones are easier to be identified.Also this method can be used in the anti-counterfeiting identification of unknown metal, and it has therefore a certain practicality.%以三种成分相异和三种成分相近的金属材料为试样,提取了10 MHz的高频超声脉冲在其内部传播时的散射信号,通过小波包变换得到散射信号在尺度空间上的能量分布,并将其作为信息防伪识别特征,再采用遗传算法优化后的BP神经网络作为分类器.结果表明,提出的方法可以成功识别成分相异和成分相近的金属材料,相比于成分相近未经热处理的金属材料,同种金属经高温热处理后更容易识别,该方法亦可用于对未知金属的防伪识别.