摘要
基于Weld-SSD算法模型和Weld-RCNN算法模型,进行焊缝图像缺陷识别算法优化,设计并实现了基于Anchor-based的Weld-RCNN算法,缺陷检测性能召回率达到84.6%,精准率70.2%。建立焊缝X射线图像缺陷识别模型并测试验证该模型的准确性。测试结果表明,算法模型在测试中表现良好,为解决焊缝X射线缺陷检测问题提供新的思路与方法。
Based on Weld-SSD and Weld-RCNN algorithm model,the defect recognition algorithm of welding seam images was optimized,and the Anchor-based Weld-RCNN algorithm was designed and implemented.The recall rate of defect detection performance reached 84.6%,and the accuracy rate was 70.2%.The welding seam X-ray image recognition model was established and tested.The results show that the model performs well and provides new ideas for solving the problem of X-ray defect detection in welding seams.
作者
许佳伟
杨亮
郝思佳
Xu Jiawei;Yang Liang;Hao Sijia(CNOOC Gas and Power Group Co.,Ltd.,Beijing,China)
出处
《科学技术创新》
2023年第18期59-62,共4页
Scientific and Technological Innovation
关键词
深度学习
焊缝
射线检测
缺陷
模型
deep learning
weld seam
radiographic test
defect
model