基于声发射信号的纳秒激光划片轮廓监测  

The Monitor of Kerf Profile Based on Acoustic Emission during Laser Scribing

作　　者：王淼峥 荣佑民 Wang Miaozheng;Rong Youmin(School of Mechanical Science and Engineering,Huazhong University of Science and technology,Wuhan 430047,Hubei,China)

机构地区：[1]华中科技大学机械科学与工程学院,湖北武汉430074

出　　处：《应用激光》2023年第3期150-155,共6页Applied Laser

基　　金：中国博士后科学基金项目(2020TQ0110)。

摘　　要：针对Al5083纳秒激光划片过程中产生沟槽和凸起两种轮廓的问题,研究了不同工艺参数下产生轮廓与映射声信号的关系。开展Al5083薄板纳秒紫外脉冲激光划片试验,观察轮廓的微观形貌,探究轮廓形成机制;采集声发射信号,小波包变换后分析声信号的差异性,并开展支持向量机分析。微观观测结果表明,凸起轮廓的成形机制包括熔融金属溅出受阻和凝固时产生的大量气孔。声信号分析结果显示,沟槽轮廓对应的小波包分解系数的方差和包络面积显著高于凸起轮廓;以小波包分解后的频谱为特征向量,添加标签后使用高斯核支持向量机分类,分类准确度达92.57%,验证了小波包变换和支持向量机的结合在基于声信号的轮廓监测中的可行性,为构建基于声发射的激光划片监测系统提供可行的技术路径。Aiming at generation of kerfs with groove and uplift profiles during laser scribing(LS)with nanosecond on Al5083,the relationships between two kinds of profiles and corresponding acoustic signals were investigated in this paper.A laser scribing experiment on Al5083 plate with nanosecond and ultraviolet laser was carried out,and microstructures of profiles was observed to explore the formation mechanism of kerfs with two different kinds of profiles,and acoustic emission(AE)signals were collected to analyze the differences.Wavelet Packet transformation(WPT)and support vector machine(SVM)analysis were performed to classify.The observation results illustrated the forming mechanism of uplift profile including the blocking of splash of molten metal and the generation of pores during solidification.AE signals analysis showed the variance and envelope area of WPT coefficients corresponding uplifts were significantly higher than those of grooves.Using WPT spectrum as feature vector after adding tags,SVM with Gaussian kernel was employed to classify.The accuracy reaching 92.57%,which verified the feasibility of WPT and SVM in the monitor of kerf profile during LS based on AE.This experiment provides a technical path for constructing LS monitor system based on AE.

关 键 词：激光划片 声发射 小波包变换 支持向量机 

分 类 号：TG485[金属学及工艺—焊接]