本文针对地下水污染源溯源研究中传统方法存在的反演不确定性、计算效率低及适用性受限等问题,提出了一种基于深度学习的污染源识别新方法。研究基于所构建的水文地质概念模型,结合地下水数值模拟技术,建立不同污染源分布及场地参数驱动下的地下水污染物迁移转化多情景数值模型,进而获得机器学习数据集,将传统反演问题转化为深度学习正演求解过程。采用正交试验设计方法生成覆盖参数空间的输入数据集,耦合地下水数值模拟与贝叶斯超参数优化的多层感知机（MLP）模型,构建了具有适用不同条件的通用性地下水污染溯源深度学习模型。针对仅有单井地下水污染监测数据不利条件,所建立的MLP模型平均确定性系数达0.79,总体平均相对误差控制在25.12%,各输出特征识别精度均超过70%。研究成果有望为单一监测点但复杂的地下水系统污染源识别提供了通用诊断模型与新的有效技术路径。
This study addresses the limitations of conventional methods in groundwater contamination source identification, including inversion uncertainty, low computational efficiency, and restricted applicability. We propose a novel deep learning-based approach for pollution source recognition. A hydrogeological conceptual model was established, integrated with multiphase flow numerical simulations to develop contaminant transport models. This framework transforms traditional inversion challenges into deep learning forward-solving processes. Through orthogonal experimental design, we generated comprehensive input datasets spanning parameter spaces. A multilayer perceptron (MLP) model was constructed by coupling multiphase flow simulations with Bayesian hyperparameter optimization, forming an adaptive deep learning framework for groundwater pollution tracing. The developed MLP model achieves an average coefficient of determination (R²) of 0.79, with overall mean relative error controlled at 25.12%. All output features demonstrate recognition accuracy exceeding 70%.This research establishes a new technical pathway for contaminant source identification in complex groundwater systems, which can be quickly applied to different contaminated site conditions.
[1] 王文兴, 童莉, 海热提. 土壤污染物来源及前沿问题[J]. 生态环境, 2005, 14(1): 1-5. DOI: 10.16258/j.cnki.1674-5906.2005.01.001.
[2] 顾文龙, 卢文喜, 张宇, 等. 基于贝叶斯推理与改进的MCMC方法反演地下水污染源释放历史[J]. 水利学报, 2016, 47(6): 772-779. DOI: 10.13243/j.cnki.slxb.20150290.
[3] Ayvaz M T.A hybrid simulation-optimization approach for solving the areal groundwater pollution source identification problems[J]. Journal of Hydrology, 2016, 538: 161-176. DOI: 10.1016/j.jhydrol.2016.04.008.
[4] Jha M, Datta B.Application of dedicated monitoring-network design for unknown pollutant-source identification based on dynamic time warping[J]. Journal of Water Resources Planning and Management, 2015, 141(11): 04015022. DOI: 10.1061/(asce)wr.1943-5452.0000513.
[5] Chakraborty A, Prakash O.Identification of clandestine groundwater pollution sources using heuristics optimization algorithms: A comparison between simulated annealing and particle swarm optimization[J]. Environmental Monitoring and Assessment, 2020, 192(12): 791. DOI: 10.1007/s10661-020-08691-7.
[6] Koupaei J A, Hosseini S M M, Maalek Ghaini F M. A new optimization algorithm based on chaotic maps and golden section search method[J]. Engineering Applications of Artificial Intelligence, 2016, 50: 201-214. DOI: 10.1016/j.engappai.2016.01.034.
[7] 高琬玉, 卢文喜, 潘紫东, 等. 基于自适应权重粒子群优化算法的地下水污染溯源辨识[J]. 中国农村水利水电, 2022(12): 1-7, 16. DOI: 10.12396/znsd.220388.
[8] 顾文龙, 卢文喜, 张宇, 等. 基于贝叶斯推理与改进的MCMC方法反演地下水污染源释放历史[J]. 水利学报, 2016, 47(6): 772-779. DOI: 10.13243/j.cnki.slxb.20150290.
[9] 潘紫东, 卢文喜, 范越, 等. 基于模拟-优化方法的地下水污染源溯源辨识[J]. 中国环境科学, 2020, 40(4): 1698-1705. DOI: 10.19674/j.cnki.issn1000-6923.2020.0189.
[10] 江思珉, 张亚力, 蔡奕, 等. 单纯形模拟退火算法反演地下水污染源强度[J]. 同济大学学报(自然科学版), 2013, 41(2): 253-257. DOI: 10.3969/j.issn.0253-374x.2013.02.017.
[11] 任加国, 龚克, 马福俊, 等. 基于BP神经网络的污染场地土壤重金属和PAHs含量预测[J]. 环境科学研究, 2021, 34(9): 2237-2247. DOI: 10.13198/j.issn.1001-6929.2021.04.22.
[12] Liu G, Zhou X, Li Q, et al.Spatial distribution prediction of soil as in a large-scale arsenic slag contaminated site based on an integrated model and multi-source environmental data[J]. Environmental Pollution, 2020, 267: 115631. DOI: 10.1016/j.envpol.2020.115631.
[13] Vesselinov V V, Alexandrov B S, O'Malley D. Nonnegative tensor factorization for contaminant source identification[J]. Journal of Contaminant Hydrology, 2019, 220: 66-97. DOI: 10.1016/j.jconhyd.2018.11.010.
[14] Chaubey J, Srivastava R.Simultaneous identification of groundwater pollution source location and release concentration using Artificial Neural Network[J]. Environmental Forensics, 2022, 23(3/4): 293-300. DOI: 10.1080/15275922.2020.1850566.
[15] Xia X M, Jiang S M, Zhou N Q, et al.Groundwater contamination source identification and high-dimensional parameter inversion using residual dense convolutional neural network[J]. Journal of Hydrology, 2023, 617: 129013. DOI: 10.1016/j.jhydrol.2022.129013.
[16] Pang M, Du E H, Zheng C M. Contaminant transport modeling and source attribution with attention-based graph neural network[J]. Water Resources Research, 2024, 60(6): e2023WR035278. DOI: 10.1029/2023WR035278.
[17] Li X, Liang G H, He B, et al.Recent advances in groundwater pollution research using machine learning from 2000 to 2023: A bibliometric analysis[J]. Environmental Research, 2025, 267: 120683. DOI: 10.1016/j.envres.2024.120683.
[18] 张涵, 郭珊珊, 胡远思. 区域地下水污染溯源与迁移转化[M]. 成都:西南交通大学出版社,2023.
[19] 王景瑞, 胡立堂. 地下水污染源识别的数学方法研究进展[J]. 水科学进展, 2017, 28(6): 943-952. DOI: 10.14042/j.cnki.32.1309.2017.06.015.
[20] Bakker M, Post V, Langevin C D, et al.Scripting MODFLOW model development using Python and FloPy[J]. Groundwater, 2016, 54(5): 733-739. DOI: 10.1111/gwat.12413.
[21] 魏亚强, 陈坚, 张铎, 等. 基于Python的地下水模拟研究进展与应用[J]. 计算机技术与发展, 2021, 31(5): 150-156. DOI: 10.3969/j.issn.1673-629X.2021.05.026.
[22] Goodfellow I, Bengio Y, Courville A.Deep learning[M]. Cambridge, MA: MIT Press, 2016:271-279.
[23] Hutter F, Hoos H H, Leyton-Brown K.Sequential model-based optimization for general algorithm configuration[C]// Learning and Intelligent Optimization. Berlin, Heidelberg: Springer Berlin Heidelberg, 2011: 507-523. DOI: 10.1007/978-3-642-25566-3_40.
[24] Rosenblatt F.The perceptron: A probabilistic model for information storage and organization in the brain[J]. Psychological Review, 1958, 65(6): 386-408. DOI: 10.1037/h0042519.
[25] 胡亮, 肖俊, 王颖. 基于多维度特征和MLP的岩体点云植被滤波方法[J]. 中国科学院大学学报, 2020, 37(3): 345-351. DOI: 10.7523/j.issn.2095-6134.2020.03.007.
[26] 张荣, 李伟平, 莫同. 深度学习研究综述[J]. 信息与控制, 2018, 47(4): 385-397, 410. DOI: 10.13976/j.cnki.xk.2018.8091.
[27] Rumelhart D E, Hinton G E, Williams R J.Learning representations by back-propagating errors[J]. Nature, 1986, 323(6088): 533-536. DOI: 10.1038/323533a0.
[28] 王碧莲,王明玉,庞云天. 典型潜水层苯系物污染羽稳定性主控因子及统计建模[J]. 地球科学,2023,48(9):3454-3465. DOI:10.3799/dqkx.2021.135.
[29] Vapnik V N .The Nature of Statistical Learning Theory[J].Springer, 1995.DOI:10.1007/978-1-4757-2440-0.
[30] Samandi V, Mukhopadhyay D.Workflow scheduling in cloud computing environment with classification ordinal optimisation using SVM[J]. International Journal of Computational Science and Engineering, 2021, 24(6): 563. DOI: 10.1504/ijcse.2021.119970.
[31] Chen W G, Gu Z L, Zou J X, et al.Analysis of furfural dissolved in transformer oil based on confocal laser Raman spectroscopy[J]. IEEE Transactions on Dielectrics and Electrical Insulation, 2016, 23(2): 915-921. DOI: 10.1109/TDEI.2015.005434.
