油田注水管网支干线泄漏诊断研究  

Research on Leakage Diagnosis of Branch Line of Oil Field Water Injection Network

作　　者：徐赋海 周军 杨海峰 孙丽 成植刚 朱家兴 XU Fuhai;ZHOU Jun;YANG Haifeng;SUN Li;CHENG Zhigang;ZHU Jiaxing(Dongrin Oil Production Plant,Dongying 257000,China;Petroleum Engineering School,Southwest Petroleum University,Chengdu 610500,China)

机构地区：[1]东辛采油厂,东营257000 [2]西南石油大学石油与天然气工程学院,成都610500

出　　处：《给水排水》2025年第2期134-139,共6页Water & Wastewater Engineering

基　　金：国家自然科学基金项目(51704253,52474084)。

摘　　要：油田注水管网发生泄漏事故时,会造成水资源的浪费。现场仅在配水间安装压力仪表,难以利用单管泄漏定位方法对支干线泄漏进行定位。基于此,建立基于卷积神经网络的泄漏预警与定位模型,通过接收输入的压力数据,对管线泄漏进行预警,基于泄漏位置与泄漏程度以及压力变化间的规律,对支干线泄漏位置进行定位。结果表明,基于卷积神经网络的泄漏预警与定位模型能针对泄漏事件做出及时预警,且定位误差在50m内。The leakage of oil field water injection network will result in waste of water resources.The pressure instrument is only installed in the water distribution room on site,so it is difficult to locate the leakage of the branch trunk by using the single pipe leakage location method.According to the above situation,the leakage warning and location model based on convolutional neural network(CNN)is established,and the pipeline leakage is predicted by receiving the input pressure data.Based on the law of leakage position,leakage degree and pressure change,the leakage position of branch trunk is located.The results show that the leakage warning and location model based on CNN can provide timely warning for leakage events,and the positioning error is within 50 m.

关 键 词：油田注水管网 卷积神经网络 泄漏预警 泄漏定位 

分 类 号：TU990[建筑科学—市政工程]