Research on construction method of water supply network partition based on deep clustering
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
维普
万方数据
在供水系统管理中,分区管理是进行管网运营、压力管理和漏损控制的有效工具,其构建方法直接影响分区效能的发挥程度.然而,大多数分区方法通常未能充分考虑管网的拓扑结构与水力状态之间的复杂相互作用,所得分区方案无法充分利用现有分区设施,在实际工程应用中面临困难.鉴于此,提出了一种基于结构性深度聚类网络(Structural Deep Clustering Network,SDCN)的供水管网分区构建方法.使用SDCN有机融合管网拓扑结构和水力状态信息进行聚类分析,根据聚类结果以及管网计量设施实际情况,对孤立节点进行归类分析,提出聚类边界调整的有效计算方法,以形成最终分区方案.与两种经典的管网聚类方法相比,该方法具有更好的管网节点聚类精度,分区结果面向工程实际需求,可以为管网分区规划与建设提供有效的技术支撑.
In the management of water supply systems,partition management serve as an ef-fective tool for network operation,pressure management,and leak control.The construction method directly affects the extent to which the efficacy of the partitioning can be realized.Howev-er,most partitioning methods often overlook the intricate interplay between the network's topolo-gical structure and hydraulic conditions,leading to the partition scheme that do not fully exploit existing partitioning infrastructure and face challenges in practical engineering applications.In re-sponse to this,this paper proposes a network partitioning strategy based on a Structural Deep Clus-tering Network(SDCN).Using SDCN to organically fuse network topology structure and hydrau-lic state information for clustering analysis;Based on the clustering results and the actual situation of the network metering facilities,an effective computational method for classifying isolated nodes and adjusting clustering boundaries is proposed to form the final metering partitioning scheme.Compared to the two classic clustering methods,this approach has better pipe network node cluste-ring accuracy,and the partition results are oriented to the actual needs of the project,which can provide valid technical support for the planning and construction of pipe network partition.
PartitioningWater distribution networkStructural deep clustering networkHy-draulic model
维普
万方数据
