焊管2025，Vol.48Issue(5)：58-66,9.DOI:10.19291/j.cnki.1001-3938.2025.05.008
基于大模型的管线钢焊缝缺陷检测研究与应用
Research and Application of Weld Defect Detection based on Large Model
程继兵 1刘占浩 2余晗 2刘明 1张毅 2赵立波 2林雪3
作者信息
- 1. 中油宝世顺(秦皇岛)钢管有限公司,河北 秦皇岛 066206
- 2. 中国石油集团宝石管业有限公司,陕西 宝鸡 721008
- 3. 华为技术有限公司,深圳 518000
- 折叠
摘要
Abstract
In the inspection process of long-distance pipeline steel pipes,X-ray inspection methods are widely used for manual assessment.However,variations in the professional expertise of inspectors result in relatively low accuracy and efficiency during inspections.To address this issue,a pipe body weld defect detection model was developed based on the Pangu large model.This model incorporates the NL-means algorithm for image denoising and leverages the pangu-cv-det-protect algorithm provided by the Huawei Cloud Modelarts platform to achieve precise detection for minor defects.By applying the AI model to the inspection images provided by a pipe factory,it was demonstrated that the model can accurately and efficiently detect defects,thereby highlighting its significant advantages in real-world production environments.In summary,this customized model is capable of rapidly and effectively identifying various weld defects,including porosity,slag inclusions,undercuts,and cracks.It not only enhances the production efficiency of steel pipes but also ensures the quality of welds in the pipe body.关键词
焊缝缺陷/AI检测模型/焊缝质量/检测精度/检测效率Key words
weld defects/AI detection model/weld quality/detection accuracy/detection efficiency分类
金属材料引用本文复制引用
程继兵,刘占浩,余晗,刘明,张毅,赵立波,林雪..基于大模型的管线钢焊缝缺陷检测研究与应用[J].焊管,2025,48(5):58-66,9.
