ARM / 싱글 파이버 레이저를 이용한 Cu-Al 용접 특성 분석에 따른 공정 최적화 및 품질 예측 알고리즘 연구

Abstract

As the necessity for secondary battery development emerges due to an increase in awareness of energy saving and environment protection, the need for not only Cu similar welding but also Cu-Al dissimilar welding increases. A welding defect of battery may cause a risk such as fire, so high-quality welding should be ensured through process optimization. However, copper welding requires a high-energy-density heat source due to its high thermal conductivity and has a strong possibility of gas absorption that easily forms pores by oxidation. In addition, Cu-Al dissimilar welding has many challenges, such as embrittlement and strength degradation due to intermetallic compound and the large differences in physical properties. Therefore, this study was conducted for analyzing welding characteristics and process optimization of Cu-Al dissimilar welding. After welding using single-mode fiber and ARM laser, the appearance and cross-section of beads were analyzed, and electrical and mechanical properties were evaluated. After establishing a real-time monitoring system and acquiring an optical signal through a photodiode, the correlation between the signal and the welding result was analyzed. Finally, a prediction model that predicts the bead size and characteristics for each experimental factor and a classification model that performs quality judgment were generated using deep learning. In addition, a model was developed by identifying the characteristics of a data file imaged with a measured signal using the CNN algorithm and learning about it. The quality of the welds was judged using the developed model.