Research on Semantic Segmentation Method of Molten Pool based on Improved YOLOv8s
- Vol. 55, Issue 11, Pages: 96-105(2025)
Received:21 November 2024,
Revised:2025-02-27,
Published:20 November 2025
DOI: 10.7512/j.issn.1001-2303.2025.11.13
移动端阅览

浏览全部资源
扫码关注微信
Received:21 November 2024,
Revised:2025-02-27,
Published:20 November 2025
移动端阅览
熔池的精准语义分割是实现焊接过程自动化质量监控的关键技术,传统识别方法受弧光干扰、动态目标定位偏差及特征融合不充分等问题制约,难以满足复杂工况下的像素级分割需求。提出了一种基于改进YOLOv8s的熔池语义分割模型。首先,在主干网络(Backbone)引入CA注意力机制,通过分离空间信息嵌入与注意力生成过程,强化模型对熔池、侧壁与焊丝位置的准确定位能力。其次,针对原PANet特征金字塔在多尺度特征融合中定位信息丢失的问题,采用双向特征金字塔(BiFPN)替换,并增设专用分割层,通过双向路径强化不同层级特征的交互融合,提升熔池边缘细节的分割能力。最后,引入EIoU损失函数替代原CIoU,通过分别优化预测框与真实框的长、宽匹配度及重叠区域评估,减少动态熔池目标的定位偏差,提升模型对熔池尺寸变化的适应性。实验结果表明,改进后的模型可实现熔池熔池及其周边区域的高精度语义分割,识别精度达99.37%,交并比为93.76%,显著提升了识别效果。该研究为焊接过程的智能化监控提供了可靠的视觉感知方案,对推动焊接自动化技术升级具有重要工程意义。
The precise semantic segmentation of the welding pool is a key technology for achieving automated quality monitoring of the welding process. Traditional recognition methods are constrained by arc light interference
dynamic target positioning deviation
and insufficient feature fusion
making it difficult to meet the pixel-level segmentation requirements under complex conditions. A semantic segmentation model of the welding pool based on an improved YOLOv8s is proposed. Firstly
a CA attention mechanism is introduced into the backbone network
and the spatial information embedding and attention generation process are separated to enhance the model's accurate positioning ability of the welding pool
side walls
and welding wire. Secondly
to address the loss of positioning information in the multi-scale feature fusion of the original PANet feature pyramid
a Bidirectional Feature Pyramid Network (BiFPN) is used as a replacement
and a dedicated segmentation layer is added. Through bidirectional paths
the interaction and fusion of features at different levels are strengthened
improving the segmentation capability of the details at the edge of the welding pool. Finally
the EIoU loss function is introduced as a replacement for the original CIoU. By optimizing the predicted and actual boxes' length
width matching degree
and overlap area evaluation separately
the positioning deviation of dynamic welding pool targets is reduced
enhancing the model's adaptability to changes in the size of the welding pool. Experimental results show that the improved model can achieve high-precision semantic segmentation of the welding pool and its surrounding areas
with a recognition accuracy of 99.37% and an intersection over union (IoU) of 93.76%
significantly improving the recognition effect. This research provides a reliable visual perception solution for intelligent monitoring of the welding process
which has significant engineering significance for promoting the upgrading of welding automation technology.
冯消冰 , 潘际銮 , 高力生 , 等 . 爬行焊接机器人在球罐自动焊接中的应用 [J]. 清华大学学报(自然科学版) , 2021 , 61 ( 10 ): 1132 - 1143 .
FENG X B , PAN J L , GAO L S , et al . Application of Crawling Welding Robot in the Automatic Welding of Spherical Tanks [J]. Journal of Tsinghua University (Science and Technology) , 2021 , 61 ( 10 ): 1132 - 1143 .
柯希林 , 王中任 , 汤宇 , 等 . 管道全位置焊接熔池监测试验研究 [J]. 机床与液压 , 2021 , 49 ( 17 ): 113 - 116 .
KE X L , WANG Z R , TANG Y , et al . Experimental Study on Weld Pool Monitoring of Pipeline All-Position Welding [J]. Machine Tool & Hydraulics , 2021 , 49 ( 17 ): 113 - 116 .
李军兆 , 孙清洁 , 张清华 , 等 . 空间多位置摆动激光填丝焊接熔池动态行为及焊缝成形 [J]. 焊接学报 , 2021 , 42 ( 10 ): 35-39+61 + 99 .
LI J Z , SUN Q J , ZHANG Q H , et al . Dynamic Behavior of Weld Pool and Weld Formation in Space Multi-Position Oscillating Laser Filler Welding [J]. Transactions of the China Welding Institution , 2021 , 42 ( 10 ): 35-39+61 + 99 .
马晓锋 , 夏攀 , 刘海生 , 等 . 全位置焊接熔池的深度学习检测方法 [J]. 机械工程学报 , 2023 , 59 ( 12 ): 272 - 283 .
MA X F , XIA P , LIU H S , et al . Deep Learning Detection Method for Weld Pool in All-Position Welding [J]. Journal of Mechanical Engineering , 2023 , 59 ( 12 ): 272 - 283 .
卢振洋 , 宫兆辉 , 闫志鸿 , 等 . 基于深度学习的TIG焊背部熔池检测和熔宽提取 [J]. 北京工业大学学报 , 2020 , 46 ( 09 ): 988 - 996 .
LU Z Y , GONG Z H , YAN Z H , et al . TIG Welding Backside Weld Pool Detection and Weld Width Extraction Based on Deep Learning [J]. Journal of Beijing University of Technology , 2020 , 46 ( 09 ): 988 - 996 .
徐东辉 , 孟范鹏 , 孙鹏 , 等 . 基于深度学习的GMAW焊接缺陷在线监测 [J]. 焊接学报 , 2024 , 45 ( 03 ): 114-119+135 - 136 .
XU D H , MENG F P , SUN P , et al . Online Monitoring of GMAW Welding Defects Based on Deep Learning [J]. Transactions of the China Welding Institution , 2024 , 45 ( 03 ): 114-119+135 - 136 .
刘天元 , 郑杭彬 , 杨长祺 , 等 . 面向激光焊接缺陷识别的可解释性深度学习方法 [J]. 航空学报 , 2022 , 43 ( 04 ): 451 - 460 .
LIU T Y , ZHENG H B , YANG C Q , et al . Explainable Deep Learning Method for Laser Welding Defect Recognition [J]. Acta Aeronautica et Astronautica Sinica , 2022 , 43 ( 04 ): 451 - 460 .
刘天元 , 鲍劲松 , 汪俊亮 , 等 . 融合时序信息的激光焊接熔透状态识别方法 [J]. 中国激光 , 2021 , 48 ( 06 ): 228 - 238 .
LIU T Y , BAO J S , WANG J L , et al . Laser Welding Penetration State Recognition Method Integrating Temporal Information [J]. Chinese Journal of Lasers , 2021 , 48 ( 06 ): 228 - 238 .
刘翰儒 , 陈桂映 , 许泽林 , 等 . 基于SERepVGG-A2的轻量化卷积神经网络在熔池状态识别中的应用 [J/OL]. 机械工程学报 , 2025 , 61 ( 03 ): 440 - 448 .
Liu Hanru , Chen Guiying , Xu Zelin , et al . Application of Light-Weight Convolutional Neural Network Based on SERepVGG-A2 in Weld Pool State Recognition [J]. Journal of Mechanical Engineering , 2025 , 61 ( 03 ): 440 - 448 .
汲生鹏 , 闫志鸿 , 刘嘉 , 等 . 基于深度学习的薄板TIG焊焊缝成形视觉检测 [J]. 热加工工艺 , 2022 , 51 ( 03 ): 115 - 121 .
JI S P , YAN Z H , LIU J , et al . Visual Inspection of Weld Formation for Thin Plate TIG Welding Based on Deep Learning [J]. Hot Working Technology , 2022 , 51 ( 03 ): 115 - 121 .
Marc-André N , Flemming J G , Dieter L , et al . Automatic melt pool recognition in X-ray radiography images from laser-molten Al alloy [J]. Journal of Materials Research and Technology , 2022 , 21 : 3502 - 3513 .
方吉米 , 王克鸿 , 黄勇 . 高速GMAW驼峰焊道形成过程熔池图像识别 [J]. 焊接学报 , 2019 , 40 ( 02 ): 42-46+162 - 163 .
FANG J M , WANG K H , HUANG Y . Image Recognition of Weld Pool during the Formation Process of High-Speed GMAW Camelback Weld Track [J]. Transactions of the China Welding Institution , 2019 , 40 ( 02 ): 42-46+162 - 163 .
冯宝 , 覃科 , 蒋志勇 . 基于L1/L2极限学习机的熔池熔透状态识别 [J]. 焊接学报 , 2018 , 39 ( 09 ): 31 - 35+130 .
FENG B , QIN K , JIANG Z Y . Penetration State Recognition of Weld Pool Based on L1/L2 Extreme Learning Machine [J]. Transactions of the China Welding Institution , 2018 , 39 ( 09 ): 31 - 35+130 .
Sohan M , Sai Ram T , Rami Reddy C V . A Review on YOLOv8 and Its Advancements [J]. ResearchGate , 2024 , 12 : 1 - 19 .
谭旭 , 赵骥 . 改进YOLOv8的汽车表面伤损实例分割模型 [J]. 计算机工程与应用 , 2024 , 60 ( 14 ): 197 - 208 .
TAN X , ZHAO J . Improved YOLOv8 Model for Instance Segmentation of Car Surface Damage [J]. Computer Engineering and Applications , 2024 , 60 ( 14 ): 197 - 208 .
王安静 , 袁巨龙 , 朱勇建 , 等 . 基于改进YOLOv8s的鼓形滚子表面缺陷检测算法 [J]. 浙江大学学报 , 2024 , 58 ( 02 ): 370 - 380+387 .
WANG A J , YUAN J L , ZHU Y J , et al . Defect Detection Algorithm for Drum-Shaped Roller Surface Based on Improved YOLOv8s [J]. Journal of Zhejiang University , 2024 , 58 ( 02 ): 370 - 380+387 .
卢子册 , 刘小芳 , 王德伟 . 基于改进YOLOv8的PCB焊点语义分割方法 [J]. 无线电工程 , 2024 , 54 ( 07 ): 1614 - 1621 .
LU Z C , LIU X F , WANG D W . Semantic Segmentation Method for PCB Solder Joints Based on Improved YOLOv8 [J]. Radio Engineering , 2024 , 54 ( 07 ): 1614 - 1621 .
周宏平 , 金寿祥 , 周磊 , 等 . 基于迁移学习与YOLOv8n的田间油茶果分类识别 [J]. 农业工程学报 , 2023 , 39 ( 20 ): 159 - 166 .
ZHOU H P , JIN S X , ZHOU L , et al . Classification and Recognition of Camellia Fruits in the Field Based on Transfer Learning and YOLOv8n [J]. Transactions of the Chinese Society of Agricultural Engineering , 2023 , 39 ( 20 ): 159 - 166 .
Mukaram Safaldin , Nizar Zaghden , Mahmoud Mejdoub . An Improved YOLOv8 to Detect Moving Objects [J]. IEEE ACCESS , 2024 , 05 , 59782 - 59806 .
相关作者
相关机构
公网安备11010802024621