Multi-view spatio-temporal graph convolutional networks model for urban drainage networks flow prediction

doi:10.11947/j.AGCS.2025.20230329

Abstract

Abstract:

Flow of urban drainage networks are critical indicators of their operational efficiency and safety. Accurately forecasting these parameters is crucial for risk mitigation, performance enhancement, and layout planning of the networks. Traditional flow forecasting methods typically overlook the complex multidimensional spatial dependencies between the flows in pipelines. This paper proposes a multi-view spatio-temporal graph convolutional networks model that considers both the spatial proximity and attribute similarity of network nodes. It constructs the nearest neighbor-based graph and the flow similarity-based graph, utilizes spatio-temporal graph convolutional networks to uncover intrinsic dependencies, and applies an attention mechanism to merge features from multiple views for enhanced flow predictions. Experiments with historical flow data from an urban drainage network confirm the superior predictive capabilities of our model, with ablation studies validating the contributions of different views.

Key words: drainage networks flow prediction, multi-view graph, spatio-temporal graph network, graph deep learning

CLC Number:

P208

Wei TU, Xiangyuan CHI, Tianhong ZHAO, Jian YANG, Shiping ZHU, Deli CHEN. Multi-view spatio-temporal graph convolutional networks model for urban drainage networks flow prediction[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(2): 334-344.

Figures/Tables 12

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Tab. 1

Fig. 7

Tab. 2

Tab. 3

Fig. 8

Fig. 9

References 28

[1]	HUANG Dong, LIU Xiuhong, JIANG Songzhu, et al. Current state and future perspectives of sewer networks in urban China[J]. Frontiers of Environmental Science & Engineering, 2018, 12(3): 2.
[2]	李清泉, 张德津, 汪驰升, 等. 动态精密工程测量技术及应用[J]. 测绘学报, 2021, 50(9): 1147-1158. DOI:. doi: 10.11947/j.AGCS.2021.20210172
	LI Qingquan, ZHANG Dejin, WANG Chisheng, et al. Technology and applications of dynamic and precise engineering surveying[J]. Acta Geodaetica et Cartographica Sinica, 2021, 50(9): 1147-1158. DOI:. doi: 10.11947/j.AGCS.2021.20210172
[3]	KHODASHENAS S R, TAJBAKHSH M. Management of urban drainage system using integrated MIKE SWMM and GIS[J]. Journal of Water Resource and Hydraulic Engineering, 2016, 5(1): 36-45.
[4]	VAN DER VOORT M, DOUGHERTY M, WATSON S. Combining Kohonen maps with ARIMA time series models to forecast traffic flow[J]. Transportation Research Part C: Emerging Technologies, 1996, 4(5): 307-318.
[5]	WANG Hao, SONG Lixiang. Water level prediction of rainwater pipe network using an SVM-based machine learning method[J]. International Journal of Pattern Recognition and Artificial Intelligence, 2020, 34(2): 2051002.
[6]	KARIMI H S, NATARAJAN B, RAMSEY C L, et al. Comparison of learning-based wastewater flow prediction methodologies for smart sewer management[J]. Journal of Hydrology, 2019, 577: 123977.
[7]	ZHANG Duo, LINDHOLM G, RATNAWEERA H. Use long short-term memory to enhance Internet of Things for combined sewer overflow monitoring[J]. Journal of Hydrology, 2018, 556: 409-418.
[8]	夏巍, 汪石, 梁祥莹. 基于改进GRU的城市供水管网流量预测研究[J]. 安徽建筑大学学报, 2023, 31(2): 51-54.
	XIA Wei, WANG Shi, LIANG Xiangying. Research on flow forecasting methods for urban water supply network based on improved GRU[J]. Journal of Anhui Jianzhu University, 2023, 31(2): 51-54.
[9]	李双宇, 张明凯, 刘艳臣, 等. 基于LSTM模型的排水系统流量预测研究[J]. 中国给水排水, 2022, 38(5): 59-64.
	LI Shuangyu, ZHANG Mingkai, LIU Yanchen, et al. Flow prediction of drainage system based on long short time memory model[J]. China Water & Wastewater, 2022, 38(5): 59-64.
[10]	NGUYEN L V, TORNYEVIADZI H M, BUI D T, et al. Predicting discharges in sewer pipes using an integrated long short-term memory and entropy A-TOPSIS modeling framework[J]. Water, 2022, 14(3): 300.
[11]	SHENG Zheng, CAI Zhikai. GAT-GRU based model for water network flow prediction[C]//Proceedings of the 9th International Conference on Water Resource and Environment. Singapore: Springer, 2024: 151-162.
[12]	REN Yibin, CHEN Huanfa, HAN Yong, et al. A hybrid integrated deep learning model for the prediction of citywide spatio-temporal flow volumes[J]. International Journal of Geographical Information Science, 2020, 34(4): 802-823.
[13]	于洋洋, 贺康杰, 武芳, 等. 面状居民地形状分类的图卷积神经网络方法[J]. 测绘学报, 2022, 51(11): 2390-2402. DOI:. doi: 10.11947/j.AGCS.2022.20210134
	YU Yangyang, HE Kangjie, WU Fang, et al. Graph convolution neural network method for shape classification of areal settlements[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(11): 2390-2402. DOI:. doi: 10.11947/j.AGCS.2022.20210134
[14]	李静, 刘海砚, 郭文月, 等. 基于深度学习的人群活动流量时空预测模型[J]. 测绘学报, 2021, 50(4): 522-531. DOI:. doi: 10.11947/j.AGCS.2021.20200230
	LI Jing, LIU Haiyan, GUO Wenyue, et al. A spatio-temporal network for human activity prediction based on deep learning[J]. Acta Geodaetica et Cartographica Sinica, 2021, 50(4): 522-531. DOI:. doi: 10.11947/j.AGCS.2021.20200230
[15]	井佩光, 田雨豆, 汪少初, 等. 基于动态扩散图卷积的交通流量预测算法[J]. 吉林大学学报(工学版), 2024, 54(6): 1582-1592.
	JING Peiguang, TIAN Yudou, WANG Shaochu, et al. Traffic flow prediction algorithm based on dynamic diffusion graph convolution[J]. Journal of Jilin University (Engineering and Technology Edition), 2024, 54(6): 1582-1592.
[16]	LIAO Ziyi, LIU Minghui, DU Bowen, et al. A temporal and spatial prediction method for urban pipeline network based on deep learning[J]. Physica A: Statistical Mechanics and Its Applications, 2022, 608: 128299.
[17]	YU Bing, YIN Haoteng, ZHU Zhanxing. Spatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting[EB/OL]. [2023-05-04]. https://arxiv.org/abs/1709.04875v4.
[18]	WU Shaofei. Spatiotemporal dynamic forecasting and analysis of regional traffic flow in urban road networks using deep learning convolutional neural network[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(2): 1607-1615.
[19]	LIU Zhichen, LIU Zhiyuan, FU Xiao. Dynamic origin-destination flow prediction using spatial-temporal graph convolution network with mobile phone data[J]. IEEE Intelligent Transportation Systems Magazine, 2022, 14(5): 147-161.
[20]	RAWAT W, WANG Zenghui. Deep convolutional neural networks for image classification: a comprehensive review[J]. Neural Computation, 2017, 29(9): 2352-2449.
[21]	DAUPHIN Y N, FAN A, AULI M, et al. Language modeling with gated convolutional networks[C]//Proceedings of the 34th International Conference on Machine Learning. Sydney: JMLR, 2017: 933-941.
[22]	KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[EB/OL]. [2023-05-04]. https://arxiv.org/abs/1609.02907v4.
[23]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2017: 1-11.
[24]	GUO Shengnan, LIN Youfang, FENG Ning, et al. Attention based spatial-temporal graph convolutional networks for traffic flow forecasting[C]//Proceedings of 2019 AAAI Conference on Artificial Intelligence. Honolulu: AAAI Press, 2019: 922-929.
[25]	NGUYEN N, QUANZ B. Temporal latent auto-encoder: a method for probabilistic multivariate time series forecasting[EB/OL]. [2023-05-04]. https://ojs.aaai.org/index.php/AAAI/article/download/17101/16908.
[26]	WU Zonghan, PAN Shirui, LONG Guodong, et al. Graph wavenet for deep spatial-temporal graph modeling[C]//Proceedings of the 28th International Joint Conference on Artificial Intelligence. Macao: ACM Press, 2019: 1907-1913.
[27]	BAI Lei, YAO Lina, LI Can, et al. Adaptive graph convolutional recurrent network for traffic forecasting[C]//Proceedings of the 34th International Conference on Neural Information Processing Systems. Vancouver: ACM Press, 2020: 17804-17815.
[28]	WUNSCH A, LIESCH T, BRODA S. Forecasting groundwater levels using nonlinear autoregressive networks with exogenous input (NARX)[J]. Journal of Hydrology, 2018, 567: 743-758.

未来时间跨度/min	MAPE	RMSE/（m³/s）
15	0.097	0.075
30	0.102	0.078
45	0.106	0.080
60	0.111	0.083

模型	MAPE				RMSE/（m³/s）
模型	15 min	30 min	45 min	60 min	15 min	30 min	45 min	60 min
ARIMA	0.176	0.238	0.258	0.276	0.107	0.141	0.153	0.165
RFR	0.146	0.161	0.175	0.190	0.090	0.095	0.101	0.110
LSTM	0.155	0.152	0.155	0.149	0.106	0.108	0.109	0.112
AutoEncoder	0.142	0.138	0.146	0.152	0.105	0.104	0.103	0.106
GWNet	0.099	0.118	0.128	0.139	0.075	0.080	0.084	0.089
AGCRN	0.095	0.104	0.118	0.118	0.081	0.085	0.087	0.090
MVSTGCN	0.097	0.102	0.106	0.111	0.075	0.078	0.080	0.083

模型	MAPE				RMSE/（m³/s）
模型	15 min	30 min	45 min	60 min	15 min	30 min	45 min	60 min
MVSTGCN	0.097	0.102	0.106	0.111	0.075	0.078	0.080	0.083
MVSTGCN-P	0.104	0.105	0.111	0.118	0.082	0.082	0.084	0.089
MVSTGCN-S	0.104	0.104	0.108	0.117	0.080	0.081	0.082	0.087