Skip to main navigation Skip to search Skip to main content

Hierarchical Self-Attention Graph Pooling Networks for Leakage Detection of Water Distribution Networks

  • Harbin Institute of Technology
  • Aerospace System Engineering Shanghai

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Leakage detection in urban water supply systems is crucial for saving water resources and ensuring social production. Unlike traditional data-driven methods, the approach introduced to leak detection in this study is based on graph neural networks, which can better explore the hidden information in the topology structure of the pipeline network itself. This article first simulates the leakage of a real pipeline network using EPANET to obtain experimental data and uses a hierarchical self-attention pooling graph neural network for graph classification. The experimental results indicated that this method has certain effectiveness in large-scale water supply networks.

Original languageEnglish
Title of host publicationProceedings of 2024 IEEE 13th Data Driven Control and Learning Systems Conference, DDCLS 2024
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages2012-2017
Number of pages6
ISBN (Electronic)9798350361674
DOIs
StatePublished - 2024
Event13th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2024 - Kaifeng, China
Duration: 17 May 202419 May 2024

Publication series

NameProceedings of 2024 IEEE 13th Data Driven Control and Learning Systems Conference, DDCLS 2024

Conference

Conference13th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2024
Country/TerritoryChina
CityKaifeng
Period17/05/2419/05/24

Keywords

  • GNN
  • Leakage Detection
  • self-attention graph pooling

Fingerprint

Dive into the research topics of 'Hierarchical Self-Attention Graph Pooling Networks for Leakage Detection of Water Distribution Networks'. Together they form a unique fingerprint.

Cite this