基于氨显色反应的LNG围护系统泄漏检测及识别方法应用  

Application of Leakage Detection and Identification Method for LNG Containment System Based on Ammonia Chromogenic Reaction

作　　者：张剑 卢志鹏 ZHANG Jian;LU Zhipeng(Hudong-Zhonghua Shipbuilding(Group)Co.,Ltd.,Shanghai 200129,China;Shanghai Shipbuilding Technology Research Institute,Shanghai 200032,China)

机构地区：[1]沪东中华造船(集团)有限公司,上海200129 [2]上海船舶工艺研究所,上海200032

出　　处：《中国海洋平台》2024年第5期103-108,共6页China offshore Platform

摘　　要：将氨显色反应方法应用在液化天然气(Liquefied Natural Gas,LNG)围护系统泄漏检测中,并探讨目标检测模型YOLOv7在氨显色图像泄漏点检测中的效果。针对LNG薄膜型围护系统泄漏检测需求,分析氨显色反应的原理,并根据该原理构建泄漏检测所需的抽注气系统、显色喷剂及图像采集终端;针对氨显色泄漏图像,将蓝色泄漏点进行标注并对数据进行MixUp数据增强;在泄漏点识别上,采用目标检测网络YOLOv7进行训练,在该数据集上平均准确度(mean Average Precision,mAP)达97.71%,可满足LNG围护系统自动化辅助检测要求。The ammonia chromogenic reaction method is applied to the leakage detection of Liquefied Natural Gas(LNG)containment systems and the effectiveness of the YOLOv7 target detection model in detecting leakage points in ammonia chromogenic images is explored.Addressing the leakage detection needs of LNG membrane containment systems,the principle of the ammonia chromogenic reaction is analyzed.Based on this principle,the necessary components for leakage detection,including the gas injection system,chromogenic spray,and image acquisition terminal,are constructed.For ammonia chromogenic leakage images,the blue leakage points are annotated and the data are enhanced using MixUp data augmentation.For leakage point identification,the YOLOv7 target detection network is trained,achieving mean Average Precision(mAP)of 97.71%on this dataset and meeting the automated auxiliary detection requirements for LNG containment systems.

关 键 词：氨显色反应 泄漏检测 LNG围护系统 目标检测 

分 类 号：TE88[石油与天然气工程—油气储运工程]