Data-driven analysis of surface roughness influence on weld quality and defect formation in laser welding of Cu–Al

Author:

Norouzian Mohammadhossein1ORCID,Amne Elahi Mahdi1,Koch Marcus2,Zaeem Reza Mahin1,Kedziora Slawomir1

Affiliation:

1. University of Luxembourg, Esch-sur-Alzette, Luxembourg

2. INM - Leibniz Institute for New Materials, Saarbrücken, Germany

Abstract

The laser welding of Cu–Al alloys for battery applications in the automotive industry presents significant challenges due to the high reflectivity of copper. Inadequate bonding and low mechanical strength may occur when the laser radiation is directed toward the copper side in an overlap configuration welding. To tackle these challenges, a laser surface treatment technique is implemented to enhance the absorption characteristics and overcome the reflective nature of the copper material. However, elevating the surface roughness and heat-energy input over threshold values leads to heightened temperature and extreme weld. This phenomenon escalates the formation of detrimental intermetallic compounds (IMC), creating defects like cracks and porosity. Metallurgical analysis, which is time-consuming and expensive, is usually used in studies to detect these phases and defects. However, to comprehensively evaluate the weld quality and discern the impact of surface structure, adopting a more innovative approach that replaces conventional cross-sectional metallography is essential. This article proposes a model based on the image feature extraction of the welds to study the effect of the laser-based structure and the other laser parameters. It can detect defects and identify the weld quality by weld classification. However, due to the complexity of the photo features, the system requires image processing and a convolutional neural network (CNN). Results show that the predictive model based on trained data can detect different weld categories and recognize unstable welds. The project aims to use a monitoring model to guarantee optimized and high-quality weld series production. To achieve this, a deeper study of the parameters and the microstructure of the weld is utilized, and the CNN model analyzes the features of 1310 pieces of weld photos with different weld parameters.

Funder

Fonds National de la Recherche Luxembourg

Publisher

SAGE Publications

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3