文献类型：期刊文献

英文题名：Quantitative Detection of Laser Welding Defective Structure Based on Feature Exaction of the Pulsed Eddy Current Signal

作者：He, Kuanfang[1];Liang, Jiahe[1];Yong, Jiangfeng[1];Shi, Wenqing[2]

机构：[1]Foshan Univ, Sch Mech Engn & Automat, Foshan 528200, Peoples R China;[2]Guangdong Ocean Univ, Coll Elect & Informat Engn, Zhanjiang 524088, Peoples R China

年份：2023

卷号：32

期号：14

起止页码：6412

外文期刊名：JOURNAL OF MATERIALS ENGINEERING AND PERFORMANCE

收录：SCI-EXPANDED(收录号：WOS:000875550000005)、、EI(收录号：20224413038947)、Scopus(收录号：2-s2.0-85140846697)、WOS

基金：This work is supported by the National Natural Science Foundation of China (62073089), Guangdong Provincial Natural Science Foundation of China (2020B1515120006), Guangdong innovation team (2020KCXTD015, 2022KCXTD029), and Guangdong Universities new information field (2021ZDZX1057), which are gratefully acknowledged.

语种：英文

外文关键词：defective structure; feature parameters; laser welding; PEC; quantitative detection

外文摘要：The laser welding structure produces defect due to the disturbance of material, technology, environment and other factors. Currently, the defect of laser welding structure is mainly judged by optical, high-speed camera, infrared thermal imaging, and x-ray method. The use of pulsed eddy current (PEC) technology can achieve a noncontact and nondestructive way for welding quality monitoring. Thus, the laser welding aluminum alloy structure with the defects of pore and crack is quantitatively detected by PEC technology. The detect system is constructed to obtain the PEC signal of laser welding seam with different defects. The feature parameters of peak value, time of peak value, fundamental wave amplitude, peak and rising curvature ratio, amplitude ratio of the fundamental and third harmonic wave, and marginal spectrum peak value of the PEC signal are calculated to represent the laser welding defect-type and size quantitatively. The defect identification model based on support vector machine (SVM) is established to identify the type and the depth of the laser welding seam defect by inputting the feature parameters. Experimental results show that the identification model achieves 95.8% accuracy of the defect type and 87.5% accuracy of the depth, which provides a method for quantitative detection of laser welding defective structure.