摘要
针对使用超声探伤方法检测金属材料时结构噪声难以消除的问题,提出将具有自适应噪声的完整集成经验模态分解(CEEMDAN)与局部熵改进的阈值算法相结合的降噪方法,对金属材料的超声信号进行降噪处理。采用CEEMDAN方法分解超声信号,得到多个本征模态函数(IMF)分量与残余分量;计算原信号的局部熵,并根据超声波的传播原理对局部熵数值进行校正,以及确定局部熵的最佳计算窗口长度,使用局部熵改进的阈值降噪算法,对含有结构噪声的IMF分量进行降噪处理;将降噪后的IMF分量与残余分量累加,获得降噪后的重构信号。通过对实测的金属材料超声信号进行试验,结果表明,该方法能够有效去除结构噪声,且最大限度地保留了缺陷信号,降噪效果明显。
Aiming at the problem of eliminating structural noise when using ultrasonic testing method to detect metallic materials is difficult,a new de-noising method combining complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN) with improved threshold algorithm and local entropy was proposed to de-nosing the ultrasonic signal of coarse-grained materials.The CEEMDAN method was used to decompose the ultrasonic signal,and the Intrinsic Mode Function (IMF) components and residual components were obtained.The local entropy of the original signal was calculated and corrected according to the propagation principle of the ultrasonic wave.And the optimal calculation window length of local entropy was determined.The threshold algorithm improved by the local entropy was used to de-nosing the IMF component containing the structural noise.Reconstructed signals after de-nosing was obtained by accumulating IMF components and residual components.Through the experiment on the measured ultrasonic signal of metallic materials,it was showed that the method could effectively remove structural noise and kept the defect signal to the maximum extent,and the effect of noise reduction was obvious.
作者
张洪
张郁
王通德
ZHANG Hong;ZHANG Yu;WANG Tongde(Jiangsu Key Laboratory of Advanced Food Manufacturing Equipment&Technology,School ofMechanical Engineering,Jiangnan University,Wuxi 214122,China;Jaingsu Nanri Hengchi ElectricEquipment Co.Ltd.,State Grid Electric Power Research Institute,Wuxi 214161,China)
出处
《探测与控制学报》
CSCD
北大核心
2019年第1期106-112,共7页
Journal of Detection & Control
关键词
超声检测
金属材料
结构噪声
信号降噪
局部熵
ultrasonic testing
metallic materials
structural noise
signal de-noise
local entropy
