一种融合时空相关性特征的高效供水管网漏损识别方法  

An Efficient Method for Identifying Leakage in Water Supply Network by Integrating Spatio-temporal Correlation Features

作　　者：胡克勇 孟欣 孙中卫 HU Ke-yong;MENG Xin;SUN Zhong-wei(School of Information and Control Engineering,Qingdao University of Technology,Qingdao 266520,China)

机构地区：[1]青岛理工大学信息与控制工程学院,山东青岛266520

出　　处：《水电能源科学》2024年第11期137-139,54,共4页Water Resources and Power

基　　金：国家自然科学基金青年基金项目(61902205);山东省自然科学基金面上项目(ZR2023MF052)。

摘　　要：供水行业的数字化转型,极大推动了数据驱动高效漏损识别方法研究,其已成为供水管网漏损控制领域研究热点。为此,提出一种融合时空相关性特征的高效供水管网漏损识别方法。该方法首先将供水管网压力时序数据与管网拓扑结构相结合来构建供水管网数据图,其蕴含数据的时空相关性特征;其次,通过改进图卷积神经网络和门控循环单元来同时提取数据图的时空相关性特征,进而构建高效供水管网漏损识别方法,实现供水管网漏损的高效识别。试验结果表明,该方法在两个供水管网上的四个评价指标结果均优于基准方法,且其具有94%以上的准确度。这将为供水管网的可持续漏损控制提供有效工具。The digital transformation of the water supply industry has greatly promoted the research on efficient data-driven leakage identification methods,which has become a hot research topic in the field of leakage control in water supply networks.Therefore,an efficient leakage identification method for water supply network is proposed,which integrates spatio-temporal correlation features.Firstly,the pressure time series data of the water supply network is combined with the network topology to construct a data graph of the water supply network,which contains the spatio-temporal correla-tion features of the data.Secondly,by improving graph convolutional neural network and gated recurrent unit to extract spatio-temporal correlation features of data graph simultaneously,an efficient method for leakage identification of water supply network is constructed to achieve efficient leakage identification of water supply network.The experimental results show that the proposed method outperforms the benchmark method in all four evaluation indicators on two water supply networks,and it has an accuracy of over 94%.Thus,it will provide effective tools for sustainable leakage control in wa-ter supply networks.

关 键 词：漏损识别 时空相关性特征 图卷积神经网络 门控循环单元 

分 类 号：TV674[水利工程—水利水电工程] TU961[建筑科学]