ST-GPINN: a spatio-temporal graph physics-informed neural network for enhanced water quality prediction in water distribution systems

Abstract Data-driven models often neglect the underlying physical principles, limiting generalization capabilities in water distribution systems (WDSs). This study presents a novel spatio-temporal graph physics-informed neural network (ST-GPINN) for water quality prediction in WDSs, integrating hydr...

Full description

Saved in:
Bibliographic Details
Main Authors: Tianwei Mu, Feiyu Duan, Baokuan Ning, Bo Zhou, Junyu Liu, Manhong Huang
Format: Article
Language:English
Published: Nature Portfolio 2025-08-01
Series:npj Clean Water
Online Access:https://doi.org/10.1038/s41545-025-00499-7
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849238206222958592
author Tianwei Mu
Feiyu Duan
Baokuan Ning
Bo Zhou
Junyu Liu
Manhong Huang
author_facet Tianwei Mu
Feiyu Duan
Baokuan Ning
Bo Zhou
Junyu Liu
Manhong Huang
author_sort Tianwei Mu
collection DOAJ
description Abstract Data-driven models often neglect the underlying physical principles, limiting generalization capabilities in water distribution systems (WDSs). This study presents a novel spatio-temporal graph physics-informed neural network (ST-GPINN) for water quality prediction in WDSs, integrating hydraulic simulations, physics-informed neural networks (PINNs), and graph neural networks (GNNs) to capture dynamics and graph-based network connectivity while approximating partial differential equations (PDEs). ST-GPINN discretizes WDSs using virtual nodes to enhance spatial granularity, employs an Encoder-Processor-Decoder architecture for predictions. Validated on Network A (a small-scale network with 9 junctions and 11 pipes) and Network B (a real large-scale WDS with 920 junctions and 1032 pipes), ST-GPINN outperforms others, achieving a MAE of 0.0073 mg/L, RMSE of 0.0121 mg/L, and R 2 of 88.91% in Network A, and a MAE of 0.008 mg/L, RMSE of 0.0098 mg/L, and R² of 98.91% in Network B. Its scalability and accuracy highlight ST-GPINN’s potential for water quality predictions.
format Article
id doaj-art-2809874ca9824f6bbafa1edf27db016e
institution Kabale University
issn 2059-7037
language English
publishDate 2025-08-01
publisher Nature Portfolio
record_format Article
series npj Clean Water
spelling doaj-art-2809874ca9824f6bbafa1edf27db016e2025-08-20T04:01:42ZengNature Portfolionpj Clean Water2059-70372025-08-018111510.1038/s41545-025-00499-7ST-GPINN: a spatio-temporal graph physics-informed neural network for enhanced water quality prediction in water distribution systemsTianwei Mu0Feiyu Duan1Baokuan Ning2Bo Zhou3Junyu Liu4Manhong Huang5School of Architecture & Civil Engineering, Shenyang University of TechnologySchool of Architecture & Civil Engineering, Shenyang University of TechnologySchool of Architecture & Civil Engineering, Shenyang University of TechnologySchool of Architecture & Civil Engineering, Shenyang University of TechnologySchool of Architecture & Civil Engineering, Shenyang University of TechnologyCollege of Environmental Science and Engineering, State Environmental Protection Engineering Center for Pollution Treatment and Control in Textile Industry, Donghua UniversityAbstract Data-driven models often neglect the underlying physical principles, limiting generalization capabilities in water distribution systems (WDSs). This study presents a novel spatio-temporal graph physics-informed neural network (ST-GPINN) for water quality prediction in WDSs, integrating hydraulic simulations, physics-informed neural networks (PINNs), and graph neural networks (GNNs) to capture dynamics and graph-based network connectivity while approximating partial differential equations (PDEs). ST-GPINN discretizes WDSs using virtual nodes to enhance spatial granularity, employs an Encoder-Processor-Decoder architecture for predictions. Validated on Network A (a small-scale network with 9 junctions and 11 pipes) and Network B (a real large-scale WDS with 920 junctions and 1032 pipes), ST-GPINN outperforms others, achieving a MAE of 0.0073 mg/L, RMSE of 0.0121 mg/L, and R 2 of 88.91% in Network A, and a MAE of 0.008 mg/L, RMSE of 0.0098 mg/L, and R² of 98.91% in Network B. Its scalability and accuracy highlight ST-GPINN’s potential for water quality predictions.https://doi.org/10.1038/s41545-025-00499-7
spellingShingle Tianwei Mu
Feiyu Duan
Baokuan Ning
Bo Zhou
Junyu Liu
Manhong Huang
ST-GPINN: a spatio-temporal graph physics-informed neural network for enhanced water quality prediction in water distribution systems
npj Clean Water
title ST-GPINN: a spatio-temporal graph physics-informed neural network for enhanced water quality prediction in water distribution systems
title_full ST-GPINN: a spatio-temporal graph physics-informed neural network for enhanced water quality prediction in water distribution systems
title_fullStr ST-GPINN: a spatio-temporal graph physics-informed neural network for enhanced water quality prediction in water distribution systems
title_full_unstemmed ST-GPINN: a spatio-temporal graph physics-informed neural network for enhanced water quality prediction in water distribution systems
title_short ST-GPINN: a spatio-temporal graph physics-informed neural network for enhanced water quality prediction in water distribution systems
title_sort st gpinn a spatio temporal graph physics informed neural network for enhanced water quality prediction in water distribution systems
url https://doi.org/10.1038/s41545-025-00499-7
work_keys_str_mv AT tianweimu stgpinnaspatiotemporalgraphphysicsinformedneuralnetworkforenhancedwaterqualitypredictioninwaterdistributionsystems
AT feiyuduan stgpinnaspatiotemporalgraphphysicsinformedneuralnetworkforenhancedwaterqualitypredictioninwaterdistributionsystems
AT baokuanning stgpinnaspatiotemporalgraphphysicsinformedneuralnetworkforenhancedwaterqualitypredictioninwaterdistributionsystems
AT bozhou stgpinnaspatiotemporalgraphphysicsinformedneuralnetworkforenhancedwaterqualitypredictioninwaterdistributionsystems
AT junyuliu stgpinnaspatiotemporalgraphphysicsinformedneuralnetworkforenhancedwaterqualitypredictioninwaterdistributionsystems
AT manhonghuang stgpinnaspatiotemporalgraphphysicsinformedneuralnetworkforenhancedwaterqualitypredictioninwaterdistributionsystems