Please wait a minute...
浙江大学学报(理学版)  2022, Vol. 49 Issue (3): 354-362    DOI: 10.3785/j.issn.1008-9497.2022.03.013
地球科学     
基于复合神经网络的多元水质指标预测模型
王昱文1,2,杜震洪1,2(),戴震1,2,刘仁义1,2,张丰1,2
1.浙江大学 浙江省资源与环境信息系统重点实验室,浙江 杭州 310028
2.浙江大学 地理信息科学研究所,浙江 杭州 310027
Multivariate water quality parameter prediction model based on hybrid neural network
Yuwen WANG1,2,Zhenhong DU1,2(),Zhen DAI1,2,Renyi LIU1,2,Feng ZHANG1,2
1.Zhejiang Provincial Key Lab of GIS,Zhejiang University,Hangzhou 310028,China
2.Department of Geographic Information Science,Zhejiang University,Hangzhou 310027,China
 全文: PDF(1744 KB)   HTML( 15 )
摘要:

长江流域在我国水资源配置体系中具有重要地位,对其进行水质预测尤为重要。基于现有研究结果,结合循环神经网络(recurrent neural network,RNN)中的门控循环单元(gate recurrent unit,GRU)模型与全连接神经网络(fully connected neural network,FCNN),提出了改进的多元水质指标预测(MWQPP)模型,并用其预测长江流域水体的pH、溶解氧(DO)、高锰酸盐指数(CODMn)、氨氮(NH3-N)。基于长江流域2011—2018年23个水质监测点7 566条原始数据,经对比实验,证明了用MWQPP模型预测得到的均方根误差(RMSE)、平均绝对误差(MAE)、平均绝对百分比误差(MAPE)和决定系数(R2)均优于传统水质预测模型,有效提升了水质预测的精度,具有较好的鲁棒性,为水质预测和流域管理提供了科学支撑。

关键词: 水质预测人工神经网络门控循环单元(GRU)全连接神经网络(FCNN)    
Abstract:

The Yangtze River basin plays an important role in Chinese water resources allocation. What proves common knowledge is that it is particularly important to predict the water quality in the Yangtze River basin. Based on the existing research, the recurrent neural network (RNN) model with gate recurrent unit (GRU) and fully connected neural network (FCNN) are combined in this study to improve a multiple water quality parameter prediction (MWQPP) model. It is proposed to predict the four water quality parameters, such as pH, dissolved oxygen (DO), permanganate index (CODMn) and ammonia nitrogen (NH3-N) in the Yangtze River basin. Based on 7 566 raw data of 23 water quality monitoring points in the Yangtze River basin from 2011 to 2018, the comparative experiments show that the root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE) and coefficient of determination (R2) obtained from the MWQPP model's prediction results are better than traditional models, such as the multiple linear regression model, the random forest model, FCNN model and LSTM model, and the MWQPP model also has better robustness than these traditional water quality prediction models. As we can say, the MWQPP model can provide scientific, reasonable and effective support for water quality assurance and water management in Yangtze River basin.

Key words: water prediction    artificial neural network    gate recurrent unit (GRU)    fully connected neural network(FCNN)
收稿日期: 2021-01-14 出版日期: 2022-05-24
CLC:  P 208  
基金资助: 国家自然科学基金资助项目(41922043);国家重点研发计划项目(2018YFB0505000)
通讯作者: 杜震洪     E-mail: duzhenhong@zju.edu.cn
作者简介: 王昱文(1996—),ORCID:https://orcid.org/0000-0003-1119-7120,女,硕士研究生,主要从事时空大数据挖掘研究.
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  
王昱文
杜震洪
戴震
刘仁义
张丰

引用本文:

王昱文, 杜震洪, 戴震, 刘仁义, 张丰. 基于复合神经网络的多元水质指标预测模型[J]. 浙江大学学报(理学版), 2022, 49(3): 354-362.

Yuwen WANG, Zhenhong DU, Zhen DAI, Renyi LIU, Feng ZHANG. Multivariate water quality parameter prediction model based on hybrid neural network. Journal of Zhejiang University (Science Edition), 2022, 49(3): 354-362.

链接本文:

https://www.zjujournals.com/sci/CN/10.3785/j.issn.1008-9497.2022.03.013        https://www.zjujournals.com/sci/CN/Y2022/V49/I3/354

图1  FCNN结构
图2  GRU结构原理
分类最小值最大值Ⅰ类Ⅱ类Ⅲ类Ⅳ类Ⅴ类
pH6.088.946 ~ 9
DO/(mg·L-12.0217.87.56532
CODMn/(mg·L-10.1010.01515203040
NH3-N/(mg·L-10.014.240.150.51.01.52.0
表1  《地表水环境质量标准GB3838—2002》中水质指数评价标准
图3  多元水质指标预测(MWQPP)模型神经网络结构
图4  训练过程MRE变化折线图
水质指标模型MAERMSEMAPER2
pHMWQPP0.1120.1710.0140.868
FCNN0.1470.2040.0190.813
LSTM0.1240.1860.0160.845
CNN-GRU0.1170.1780.0140.851
RF0.1230.1810.0160.862
MLR0.1160.1690.0150.878
DOMWQPP0.4390.6100.0580.872
FCNN0.5170.6970.0670.832
LSTM0.4420.6150.0590.869
CNN-GRU0.4480.6150.0630.864
RF0.5070.7130.0680.825
MLR0.4890.6970.0650.832
CODMnMWQPP0.3140.4620.1660.744
FCNN0.3690.5110.1900.687
LSTM0.3250.4730.1740.731
CNN-GRU0.3220.4690.1700.739
RF0.4200.6250.2290.617
MLR0.3750.5850.2120.665
NH3-NMWQPP0.0540.1070.2820.808
FCNN0.0900.1600.5250.579
LSTM0.0580.1130.3100.788
CNN-GRU0.0580.1110.3030.793
RF0.0680.1310.3070.721
MLR0.0630.1270.2850.741
表2  水质指标预测模型结果评价与比较
图5  某点位水质指标各模型预测值与真实值比较
图6  某点位水质指标逐年预测MAPE比较
  
图7  多点位水质指标预测MAPE 比较
  
水质指标噪声比例/%MAPE
MWQPPFCNNLSTMCNN-GRURFMLR
pH00.0140.0190.0160.0140.0160.015
200.0140.0250.0170.0140.0170.015
400.0140.0280.0170.0150.0170.016
600.0140.0320.0180.0150.0170.016
DO00.0580.0670.0590.0630.0680.065
200.0580.070.0610.0640.0680.065
400.0580.0740.0630.0660.0670.065
600.0590.0850.0640.0660.0670.065
CODMn00.1660.190.1740.170.2290.212
200.1650.1750.1720.1680.2210.212
400.1650.1670.1690.1660.2280.211
600.1640.1480.1640.1660.220.205
NH3-N00.2820.5250.310.3030.3070.285
200.2810.4650.30.2990.3110.282
400.2810.4490.2940.2970.3150.274
600.2790.3640.2990.2960.3160.266
表3  不同噪声下各模型预测结果的MAPE对比
1 BERNHARD L, HØJSTRUP J. Evaluation of the wind-resource estimation program WAsP for offshore applications[J]. Journal of Wind Engineering and Industrial Aerodynamics, 2001, 89(3/4): 271-291. DOI:10.1016/S0167-6105(00)00082-9
doi: 10.1016/S0167-6105(00)00082-9
2 MARTIN N, MCEACHERN P M, YU T, et al. Model development for prediction and mitigation of dissolved oxygen sags in the Athabasca River, Canada[J]. Science of the Total Environment, 2013, 443: 403-412. DOI:10.1016/j.scitotenv.2012. 10.030
doi: 10.1016/j.scitotenv.2012. 10.030
3 BAUWE A, ECKHARDT K U, LENNARTZ B. Predicting dissolved reactive phosphorus in tile-drained catchments using a modified SWAT model[J]. Ecohydrology and Hydrobiology, 2019, 19(2): 198-209. DOI:10.1016/j.ecohyd.2019.03.003
doi: 10.1016/j.ecohyd.2019.03.003
4 秦文虎, 付亚涛. 基于向量自回归模型的水质异常检测研究[J]. 安全与环境学报, 2018, 18(4): 1560-1563. DOI:10.13637/j.issn.1009-6094.2018.04.056
QIN W H, FU Y T. Water quality abnormity inspection and detection via the vector auto-regressive model[J]. Journal of Safety and Environment, 2018, 18(4): 1560-1563. DOI:10.13637/j.issn.1009-6094.2018.04.056
doi: 10.13637/j.issn.1009-6094.2018.04.056
5 CHEN W, XU H L, CHEN Z S, et al. A novel method for time series prediction based on error decomposition and nonlinear combination of forecasters[J]. Neurocomputing, 2021, 426: 85-103. DOI:10.1016/j.neucom.2020.10.048
doi: 10.1016/j.neucom.2020.10.048
6 马晋, 何鹏, 杨庆, 等. 基于回归分析的地下水污染预警模型[J]. 环境工程, 2019, 37(10): 211-215. DOI:10.13205/j.hjgc.201910036
MA J, HE P, YANG Q, et al. Early warning model for groundwater pollution based on regression analysis[J]. Environmental Engineering, 2019, 37(10): 211-215. DOI:10.13205/j.hjgc.201910036
doi: 10.13205/j.hjgc.201910036
7 LI X, CHENG Z, YU Q, et al. Water-quality prediction using multimodal support vector regression: Case study of Jialing River, China[J]. Journal of Environmental Engineering, 2017, 143(10): 04017070. doi:10.1061/(asce)ee.1943-7870.0001272
doi: 10.1061/(asce)ee.1943-7870.0001272
8 白云, 李勇. 基于VMD-LSSVR的河水水质预测方法[J]. 安全与环境学报, 2020, 20(3): 1162-1168. DOI:10.13637/j.issn.1009-6094.2019.0727
BAI Y, LI Y. A prediction model of the river water quality based on the VMD-LSSVR[J]. Journal of Safety and Environment, 2020, 20(3): 1162-1168. DOI:10.13637/j.issn.1009-6094.2019.0727
doi: 10.13637/j.issn.1009-6094.2019.0727
9 薛同来, 赵冬晖, 韩菲. 基于GA优化的SVR水质预测模型研究[J]. 环境工程, 2020, 38(3): 123-127, 128. DOI:10.13205/j.hjgc.202003021
XUE T L, ZHAO D H, HAN F. SVR water quality prediction model based on GA optimization[J]. Environmental Engineering, 2020, 38(3): 123-127, 128. DOI:10.13205/j.hjgc.202003021
doi: 10.13205/j.hjgc.202003021
10 王晓萍, 孙继洋, 金鑫. 基于BP神经网络的钱塘江水质指标的预测[J]. 浙江大学学报(工学版), 2007, 41(2): 361-364. DOI:10.3785/j.issn.1008-973X.2007.02.037
WANG X P, SUN J Y, JIN X. Prediction of water quality index in Qiantang River based on BP neural network model[J]. Journal of Zhejiang University(Engineering Science), 2007, 41(2): 361-364. DOI:10.3785/j.issn.1008-973X.2007.02.037
doi: 10.3785/j.issn.1008-973X.2007.02.037
11 宓云軿, 王晓萍, 金鑫. 基于机器学习的水质COD预测方法[J]. 浙江大学学报(工学版), 2008, 42(5): 790-793. DOI:10.3785/j.issn.1008-973X.2008.05.013
MI Y P, WANG X P, JIN X. Water COD prediction based on machine learning[J]. Journal of Zhejiang University(Engineering Science), 2008,42(5): 790-793. DOI:10.3785/j.issn.1008-973X.2008.05.013
doi: 10.3785/j.issn.1008-973X.2008.05.013
12 SAMSUDIN M S, AZID A, KHALIT S I, et al. Comparison of prediction model using spatial discriminant analysis for marine water quality index in Mangrove Estuarine Zones[J]. Marine Pollution Bulletin, 2019, 141: 472-481. DOI:10.1016/j.marpolbul.2019.02.045
doi: 10.1016/j.marpolbul.2019.02.045
13 RAJAEE T, KHANI S, RAVANSALAR M. Artificial intelligence-based single and hybrid models for prediction of water quality in rivers: A review[J]. Chemometrics and Intelligent Laboratory Systems, 2020, 200: 103978. DOI:10.1016/j.chemolab.2020. 103978
doi: 10.1016/j.chemolab.2020. 103978
14 TA X X, WEI Y. Research on a dissolved oxygen prediction method for recirculating aquaculture systems based on a convolution neural network[J]. Computers and Electronics in Agriculture, 2018, 145: 302-310. DOI:10.1016/j.compag.2017.12.037
doi: 10.1016/j.compag.2017.12.037
15 NOORI N, KALIN L, ISIK S. Water quality prediction using SWAT-ANN coupled approach[J]. Journal of Hydrology, 2020, 590: 125220. DOI:10. 1016/j.jhydrol.2020.125220
doi: 10. 1016/j.jhydrol.2020.125220
16 WANG Z, MAN Y, HU Y, et al. A deep learning based dynamic COD prediction model for urban sewage[J]. Environmental Science (Water Research & Technology), 2019, 5(12): 2210-2218. DOI:10.1039/C9EW00505F
doi: 10.1039/C9EW00505F
17 TU J C, YANG X Q, CHEN C B, et al. Water quality prediction model based on GRU hybrid network[C]// 2019 Chinese Automation Congress (CAC). Hangzhou: CAC, 2019: 1893-1898. DOI:10.1109/CAC48633.2019. 8996847
doi: 10.1109/CAC48633.2019. 8996847
18 HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780. DOI:10.1162/neco.1997.9.8.1735
doi: 10.1162/neco.1997.9.8.1735
19 CHO K, MERRIËNBOER B VAN, BAHDANAU D, et al. On the properties of neural machine translation: Encoder-decoder approaches[EB/OL].[2014-10-07] . DOI:10.48550/arXiv. 1409.1259
doi: 10.48550/arXiv. 1409.1259
20 BILALI A E, TALEB A, BROUZIYNE Y. Groundwater quality forecasting using machine learning algorithms for irrigation purposes[J]. Agricultural Water Management, 2021, 245: 106625. DOI:10.1016/j.agwat.2020.106625
doi: 10.1016/j.agwat.2020.106625
21 刘世存, 杨薇, 田凯, 等. 基于多层全连接神经网络的白洋淀水质预测[J]. 农业环境科学学报, 2020, 39(6): 1283-1292. DOI:10.11654/jaes.2020-0288
LIU S C, YANG W, TIAN K, et al. Water quality forecasting based on multilayer fully connected neural network for Baiyangdian Lake[J]. Journal of Agro-Environment Science, 2020, 39(6): 1283-1292. DOI:10.11654/jaes.2020-0288
doi: 10.11654/jaes.2020-0288
[1] 郭伟强. 用人工神经网络方法解析色谱重叠峰[J]. 浙江大学学报(理学版), 1998, 25(3): 62-65.