Please wait a minute...
浙江大学学报(工学版)
浙江大学学报(工学版)  2021, Vol. 55 Issue (4): 601-607    DOI: 10.3785/j.issn.1008-973X.2021.04.001
计算机技术、电信技术     
基于图神经网络的地表水水质预测模型
许佳辉1(),王敬昌2,陈岭1,*(),吴勇2
1. 浙江大学 计算机科学与技术学院，浙江 杭州 310027
2. 浙江鸿程计算机系统有限公司，浙江 杭州 310009
Surface water quality prediction model based on graph neural network
Jia-hui XU1(),Jing-chang WANG2,Ling CHEN1,*(),Yong WU2
1. College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China
2. Zhejiang Hongcheng Computer Systems Limited Company, Hangzhou 310009, China
 全文: PDF(750 KB)   HTML
摘要：

针对水质数据在时间和空间维度上的复杂依赖关系，提出基于图神经网络（GNN）的地表水水质预测模型. 该模型采用GNN建模地表水水质监测站点在空间上的复杂依赖关系，使用长短时记忆网络（LSTM）建模水质指标序列在时间上的复杂依赖关系，将编码结果输入到解码器中得到预测输出. 实验结果表明，与时间序列分析方法、通用回归方法和一般深度学习方法相比，该模型能够实现23.3%、26.6%和14.8%的性能提升.
关键词： 水质预测;  图神经网络（GNN）;  深度神经网络;  长短时记忆网络（LSTM）;  深度学习    
Abstract:

A surface water quality prediction model based on graph neural network (GNN) was proposed to solve the problem that water quality data has complex dependencies in both temporal and spatial dimensions. GNN was utilized to model the complex spatial dependencies of monitoring stations, and long short-term memory (LSTM) was used to model the complex temporal dependencies of historical water quality sequences. Then the encoded vector was input into the decoder to get the water quality prediction output. The experimental results show that the model can achieve 23.3%, 26.6% and 14.8% performance improvements compared with time series analysis methods, general regression methods and existing deep learning methods.
Key words: water quality prediction    graph neural network (GNN)    deep neural network    long short-term memory (LSTM)    deep learning
收稿日期: 2021-01-19 出版日期: 2021-05-07
CLC:  TP 319  
基金资助: “十三五”水体污染控制与治理科技重大专项资助项目（2018ZX07208-009）；中央高校基本科研业务费专项资金资助项目（2020QNA5017）
通讯作者: 陈岭     E-mail: xujiahui19@zju.edu.cn;lingchen@cs.zju.edu.cn
作者简介: 许佳辉（1995—），男，硕士生，从事时空数据挖掘的研究. orcid.org/0000-0001-6010-8433. E-mail： xujiahui19@zju.edu.cn
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章　　
许佳辉
王敬昌
陈岭
吴勇

引用本文:

许佳辉,王敬昌,陈岭,吴勇. 基于图神经网络的地表水水质预测模型[J]. 浙江大学学报(工学版), 2021, 55(4): 601-607.

Jia-hui XU,Jing-chang WANG,Ling CHEN,Yong WU. Surface water quality prediction model based on graph neural network. Journal of ZheJiang University (Engineering Science), 2021, 55(4): 601-607.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2021.04.001    或    http://www.zjujournals.com/eng/CN/Y2021/V55/I4/601

图 1  地表水水质预测模型框架图
水质指标 有效值
PH值 0~14
氨氮质量浓度 0~3 mg/L
总磷质量浓度 0~0.6 mg/L
高锰酸盐指数 0~22.5 mg/L
溶解氧质量浓度 0~12 mg/L
表 1  站点监测数据指标及有效值范围
气象指标 有效值
气温 ?20~50 °C
气压 900~1100 hPa
湿度 0~100%
降雨量 0~60 mm
表 2  气象数据数据指标及有效值范围
图 2  站点图的构建
图 3  杭嘉湖地区水质监测站点分布图
图 4  距离阈值对均方误差的影响
方法 PH监测值 高锰酸盐指数 溶解氧质量浓度
MAE MRE MAE MRE MAE MRE
本文模型 0.099 1.4% 0.468 10.7% 0.478 7.5%
w/o wea 0.109 1.6% 0.478 10.9% 0.484 7.6%
w/o water 0.123 1.7% 0.608 13.3% 0.662 10.7%
表 3  变体实验结果
方法 $L=1$ $L=3$ $L=5$ $L=7$ $L=9$
MAE MRE MAE MRE MAE MRE MAE MRE MAE MRE
本文模型 0.081 1.2% 0.090 1.3% 0.096 1.4% 0.097 1.4% 0.110 1.6%
SVR 0.101 1.5% 0.125 1.8% 0.135 1.9% 0.144 2.0% 0.167 2.2%
ARIMA 0.079 1.1% 0.143 2.0% 0.138 1.9% 0.135 1.8% 0.167 2.2%
LSTM 0.089 1.3% 0.125 1.8% 0.110 1.6% 0.105 1.6% 0.126 1.8%
表 4  PH监测值的预测结果
方法 $L=1$ $L=3$ $L=5$ $L=7$ $L=9$
MAE MRE MAE MRE MAE MRE MAE MRE MAE MRE
本文模型 0.410 8.2% 0.455 10.5% 0.474 10.8% 0.497 11.1% 0.520 11.4%
SVR 0.624 13.7% 0.673 14.6% 0.680 14.8% 0.671 14.6% 0.681 14.8%
ARIMA 0.571 12.9% 0.712 15.5% 0.695 15.1% 0.743 16.0% 0.850 17.7%
LSTM 0.521 11.4% 0.532 11.7% 0.589 13.2% 0.641 14.0% 0.674 14.7%
表 5  高锰酸盐指数的预测结果
方法 $L=1$ $L=3$ $L=5$ $L=7$ $L=9$
MAE MRE MAE MRE MAE MRE MAE MRE MAE MRE
本文模型 0.354 5.7% 0.451 7.1% 0.466 7.3% 0.517 8.3% 0.582 9.2%
SVR 0.525 8.4% 0.731 11.5% 0.803 12.0% 0.897 13.9% 0.876 13.7%
ARIMA 0.501 8.0% 0.818 12.3% 0.802 12.0% 0.865 13.6% 0.891 13.8%
LSTM 0.467 7.3% 0.672 10.9% 0.675 10.8% 0.742 11.7% 0.839 12.7%
表 6  溶解氧质量浓度的预测结果
1 LIANG J, YANG Q, SUN T, et al MIKE 11 model-based water quality model as a tool for the evaluation of water quality management plans[J]. Journal of Water Supply: Research and Technology-AQUA, 2015, 64 (6): 708- 718
doi: 10.2166/aqua.2015.048
2 GONG R, XU L, WANG D, et al Water quality modeling for a typical urban lake based on the EFDC model[J]. Environmental Modeling and Assessment, 2016, 21 (5): 643- 655
doi: 10.1007/s10666-016-9519-1
3 FARUK D O A hybrid neural network and ARIMA model for water quality time series prediction[J]. Engineering Applications of Artificial Intelligence, 2010, 23 (4): 586- 594
doi: 10.1016/j.engappai.2009.09.015
4 PARMAR K S, BHARDWAJ R Water quality management using statistical analysis and time-series prediction model[J]. Applied Water Science, 2014, 4 (4): 425- 434
doi: 10.1007/s13201-014-0159-9
5 LIU S, TAI H, DING Q, et al A hybrid approach of support vector regression with genetic algorithm optimization for aquaculture water quality prediction[J]. Mathematical and Computer Modelling, 2013, 58 (3): 458- 465
6 SINGH K P, BASANT N, GUPTA S Support vector machines in water quality management[J]. Analytica Chimica Acta, 2012, 703 (2): 152- 162
7 CHENG S, ZHANG S, LI L, et al Water quality monitoring method based on TLD 3D Fish tracking and XGBoost[J]. Mathematical Problems in Engineering, 2018, 2018 (7): 1- 12
8 BUI D T, KHOSRAVI K, TIEFENBACHER J, et al Improving prediction of water quality indices using novel hybrid machine-learning algorithms[J]. Science of the Total Environment, 2020, 721 (15): 137612
9 LI L, JIANG P, XU H, et al Water quality prediction based on recurrent neural network and improved evidence theory: a case study of Qiantang River, China[J]. Environmental Science and Pollution Research, 2019, 26 (4): 19879- 19896
10 WANG Y, ZHOU J, CHEN K, et al. Water quality prediction method based on LSTM neural network [C]// International Conference on Intelligent Systems and Knowledge Engineering. Nanjing: IEEE, 2017.
11 LIU P, WANG J, CHEN K, et al Analysis and prediction of water quality using LSTM deep neural networks in IoT environment[J]. Sustainability, 2019, 11 (4): 2058
12 KIPF T, WELLING M. Semi-supervised classification with graph convolutional networks [C]// International Conference on Learning Representations. Toulon: [s. n.], 2017.
13 HOCHREITER S, SCHMIDHUBER J Long short-term memory[J]. Neural Computation, 1997, 9 (8): 1735- 1780
doi: 10.1162/neco.1997.9.8.1735
14 中华人民共和国生态环境部. 地表水环境质量标准: GB 3838—2002 [S]. 北京: 中国环境出版集团, 2002.
15 PASZKE A, GROSS S, MASSA F, et al. Pytorch: an imperative style, high-performance deep learning library [C]// Advances in Neural Information Processing Systems. Vancouver: Curran Associates, 2019.
16 FEY M, LENSSEN J E. Fast graph representation learning with PyTorch geometric [C]// International Conference on Learning Representations RLGM Workshop. New Orleans: [s. n.], 2019.
[1] 王虹力,郭斌,刘思聪,刘佳琪,仵允港,於志文. 边端融合的终端情境自适应深度感知模型[J]. 浙江大学学报(工学版), 2021, 55(4): 626-638.
[2] 张腾,蒋鑫龙,陈益强,陈前,米涛免,陈彪. 基于腕部姿态的帕金森病用药后开-关期检测[J]. 浙江大学学报(工学版), 2021, 55(4): 639-647.
[3] 毛奕喆,龚国芳,周星海,王飞. 基于马尔可夫过程和深度神经网络的TBM围岩识别[J]. 浙江大学学报(工学版), 2021, 55(3): 448-454.
[4] 徐利锋,黄海帆,丁维龙,范玉雷. 基于改进DenseNet的水果小目标检测[J]. 浙江大学学报(工学版), 2021, 55(2): 377-385.
[5] 陈世达,刘强,韩亮. 降低分布式训练通信的梯度稀疏压缩方法[J]. 浙江大学学报(工学版), 2021, 55(2): 386-394.
[6] 陈纬奇,王敬昌,陈岭,杨勇勤,吴勇. 基于深度神经网络的多因素感知终端换机预测模型[J]. 浙江大学学报(工学版), 2021, 55(1): 109-115.
[7] 许豪灿,李基拓,陆国栋. 由LeNet-5从单张着装图像重建三维人体[J]. 浙江大学学报(工学版), 2021, 55(1): 153-161.
[8] 黄毅鹏,胡冀苏,钱旭升,周志勇,赵文露,马麒,沈钧康,戴亚康. SE-Mask-RCNN：多参数MRI前列腺癌分割方法[J]. 浙江大学学报(工学版), 2021, 55(1): 203-212.
[9] 陈巧红,陈翊,李文书,贾宇波. 多尺度SE-Xception服装图像分类[J]. 浙江大学学报(工学版), 2020, 54(9): 1727-1735.
[10] 郑浦,白宏阳,李伟,郭宏伟. 复杂背景下的小目标检测算法[J]. 浙江大学学报(工学版), 2020, 54(9): 1777-1784.
[11] 周登文,田金月,马路遥,孙秀秀. 基于多级特征并联的轻量级图像语义分割[J]. 浙江大学学报(工学版), 2020, 54(8): 1516-1524.
[12] 明涛,王丹,郭继昌,李锵. 基于多尺度通道重校准的乳腺癌病理图像分类[J]. 浙江大学学报(工学版), 2020, 54(7): 1289-1297.
[13] 闫旭,范晓亮,郑传潘,臧彧,王程,程明,陈龙彪. 基于图卷积神经网络的城市交通态势预测算法[J]. 浙江大学学报(工学版), 2020, 54(6): 1147-1155.
[14] 江金生,任浩然,李瀚野. 基于卷积自编码器的地震数据处理[J]. 浙江大学学报(工学版), 2020, 54(5): 978-984.
[15] 汪周飞,袁伟娜. 基于深度学习的多载波系统信道估计与检测[J]. 浙江大学学报(工学版), 2020, 54(4): 732-738.