基于YOLOv5的管道漏磁内检测环焊缝识别方法研究  

Research on Identification Method of Girth Weld in Pipeline MagneticFlux Leakage Internal Detection Based on YOLOv5

作　　者：孟祥来 吕岩 李彦春 蒲晓晨 李春晖 杨椀迪 Meng Xianglai;Lv Yan;Li Yanchun;Pu Xiaochen;Li Chunhui;Yang Wandi(China Oil Pipeline Testing Technology Co.,LTD.,Langfang 065000,China)

机构地区：[1]中油管道检测技术有限责任公司,河北廊坊065000

出　　处：《黑龙江科学》2025年第8期38-41,共4页Heilongjiang Science

基　　金：国家重点研发计划项目(2023YFF0615101)资助。

摘　　要：管道漏磁内检测技术是确保长输油气管道安全稳定运行的关键技术之一,借助人工智能中的深度学习技术可实现自动识别管道漏磁内检测数据,利用YOLOv5目标检测算法,对管道漏磁内检测数据图片进行训练,训练后的模型能够自动标记出漏磁内检测数据图片中的环焊缝。试验表明,基于YOLOv5算法可以实现管道中环焊缝的自动识别,在合适的训练条件下,环焊缝识别数量的精度可达99.85%,说明使用该方法对漏磁内检测数据中的环焊缝特征进行自动识别是可行的,效果较好。As one of the important methods for safety detection of long-distance oil and gas pipelines,pipeline magnetic flux leakage detection technology plays an important role in guaranteeing the safe operation of pipelines.With the help of the deep learning technology of artificial intelligence,the automatic identification of pipeline magnetic flux leakage detection data can be realized,when object detection algorithm of YOLOv5 is used to train images of the pipeline leakage detection data.The trained model can automatically mark the girth weld in images of the pipeline magnetic flux leakage detection data.Through the test,based on the YOLOv5 algorithm,it can realize the automatic identification of magnetic flux leakage signals of girth weld in the pipeline.Under suitable training conditions,the accuracy of identifying the number of girth weld can reaches 99.85%,which indicates that using this method is a good choice to automatically recognize the characteristics of the girth weld from the magnetic flux leakage detection data.

关 键 词：管道漏磁内检测 环焊缝识别 深度学习 目标检测 YOLOv5 

分 类 号：TE973.6[石油与天然气工程—石油机械设备] TP183[自动化与计算机技术—控制理论与控制工程]