| 955 | 2 | 1382 |
| 下载次数 | 被引频次 | 阅读次数 |
【目的】在水文地质与工程地质研究中，传统机理数值模型在模拟复杂物理过程时存在建模精度低、不确定性等问题，机器学习模型则存在数据需求量大和可解释性差的不足。物理信息神经网络(Physics-Informed Neural Networks, PINNs)作为一种结合物理定律和机器学习的新方法，能够为解决上述问题提供可行的方案。【方法】首先，通过整理近四年文献，系统梳理机理数值模型、机器学习模型以及机理-学习耦合模型在水文地质与工程地质领域的研究现状；其次，深入分析PINNs在该领域的最新应用；最后，阐述了PINNs在水文地质与工程地质领域发展中存在的问题，并对其今后的发展给出相关建议。【结果】研究发现，在水文地质与工程地质领域，PINNs部分解决了数值模型和机器学习模型中存在的数据稀缺、可解释性差和泛化性不足的问题，拥有广阔的应用前景。今后需要进一步解决其在鲁棒性、自适应权重分配和初边界条件处理方面的问题，深入挖掘其潜力。【结论】在未来研究中，建议耦合生成式模型或强化学习等模型，减少因数据质量和噪声对模型的影响，提高PINNs的鲁棒性；使用自适应学习算法和动态权重平衡机制，平衡损失函数各项权重，使PINNs模型输出矩阵满足正交条件，提高PINNs模型的计算效率；综合实际情况，选择优化激活函数、约束方式等，使PINNs模型收敛速度更快，结果更为精准。
Abstract：[Objective] In the research on hydrogeology and engineering geology, traditional mechanism-based numerical models often exhibit issues such as low modeling accuracy and high uncertainty when simulating complex physical processes, while machine learning models are limited by their substantial data demands and poor interpretability. Physics-Informed Neural Networks(PINNs), as a new computational method that combines physical laws and machine learning, can provide a feasible solution to the above-mentioned problems. [Methods] Firstly, recent literature from the past four years is systematically reviewed to summarize the current research status of mechanism-based numerical models, machine learning models, and mechanism-learning coupled models in the fields of hydrogeology and engineering geology. Secondly, an in-depth analysis of PINNs' latest applications in these fields is conducted. Finally, the existing limitations of PINNs in the fields of hydrogeology and engineering geology are expounded, and recommendations for future development are proposed. [Results] It is found that PINNs have partially addressed issues in numerical models and machine learning models, such as data scarcity, poor interpretability, and insufficient generalizability, demonstrating broad application prospects in hydrogeology and engineering geology. Future efforts should focus on resolving existing problems in robustness, adaptive weight allocation, and initial and boundary condition processing to further tap into their potential. [Conclusion] In future research, the following recommendations for further development are proposed. Generative models or reinforcement learning should be coupled to reduce the impact of data quality and noise on the model, thereby enhancing robustness. Adaptive learning algorithms and dynamic weight balancing mechanisms should be employed to balance the weights of each term in the loss function and ensure the output matrix of the PINNs model satisfies orthogonal conditions, thereby improving computational efficiency. Considering the actual situation, methods such as the activation function and constraint methods should be optimized for selection to achieve faster convergence and higher accuracy in PINN modeling.
[1] 王浩，陆垂裕，秦大庸，等.地下水数值计算与应用研究进展综述[J].地学前缘，2010,17(6):1-12.WANG H,LU C Y,QIN D Y,et al.Advances in method and application of groundwater numerical simulation[J].Earth Science Frontiers,2010,17(6):1-12.
[2] 郎宝山，范显利，薄浩，等.考虑各向异性的砂岩地热储层提高数值模拟精度方法[J].石油钻采工艺，2024,46(3):309-316.LANG Baoshan,FAN Xianli,BO Hao,et al.A method for improving anisotropic numerical simulation accuracy of sandstone geothermal reservoirs[J].Oil Drilling & Production Technology,2024,46(3):309-316.
[3] 裴韬，刘亚溪，郭思慧，等.地理大数据挖掘的本质[J].地理学报，2019,74(3):586-598.PEI T,LIU Y X,GUO S H,et al.Principle of big geodata mining[J].Acta Geographica Sinica,2019,74(3):586-598.
[4] BERGEN K J,JOHNSON P A,DE HOOP M V,et al.Machine learning for data-driven discovery in solid Earth geoscience[J].Science,2019,363(6433):eaau0323.
[5] 王雪飞，高科超，张兴华，等.试井数据监督式机器学习方法预测变质岩潜山储层产量[J].石油钻采工艺，2024,46(4):443-454.WANG Xuefei,GAO Kechao,ZHANG Xinghua,et al.A supervised machine learning method based on well test data to predict reservoir production in buried-hill metamorphic rock[J].Oil Drilling & Production Technology,2024,46(4):443-454.
[6] AHMADI A,OLYAEI M,HEYDARI Z,et al.Groundwater level modeling with machine learning:a systematic review and meta-analysis[J].Water,2022,14(6):949.
[7] 沈焕锋，张良培.地球表层特征参量反演与模拟的机理-学习耦合范式[J].中国科学：地球科学，2023,53(3):546-560.SHEN H F,ZHANG L P.Mechanism-learning coupling paradigms for parameter inversion and simulation in earth surface systems[J].Science China Earth Sciences,2023,53(3):546-560.
[8] RASHT-BEHESHT M,HUBER C,SHUKLA K,et al.Physics-Informed Neural Networks (PINNs) for wave propagation and full waveform inversions[J].Journal of Geophysical Research:Solid Earth,2022,127(5):e2021JB023120.
[9] ZHANG E,DAO M,KARNIADAKIS E,et al.Analyses of internal structures and defects in materials using Physics-Informed Neural Networks[J].Science advances,2022,8(7):eabk0644.
[10] TENDERINI R,PAGANI S,QUARTERONI A,et al.PDE-aware deep learning for inverse problems in cardiac electrophysiology[J].SIAM Journal on Scientific Computing,2022,44(3):B605-B639.
[11] SARABIAN M,BABAEE H,LAKSARI K.Physics-Informed Neural Networks for brain hemodynamic predictions using medical imaging[J].IEEE transactions on medical imaging,2022,41(9):2285-2303.
[12] FANG Z,ZHAN J.Deep physical informed neural networks for metamaterial design[J].IEEE Access,2019,8:24506-24513.
[13] 王燕嘉，邱璐，朱剑琴.基于物理神经网络的导热问题无网格计算方法[J].航空动力学报，2023,38(7):1626-1636.WANG Y J,QIU L,ZHU J Q.Meshless computing method for thermal conductivity problem based on physical-informed neural networks[J].Journal of Aerospace Power,2023,38(7):1626-1636.
[14] RAAZIA S,DAR Q.A numerical model of groundwater flow in Karewa-Alluvium aquifers of NW Indian Himalayan Region[J].Modeling Earth Systems and Environment,2022,8(1):871-882.
[15] VU D,NI F,LI C,et al.Predictions of groundwater vulnerability and sustainability by an integrated index-overlay method and physical-based numerical model[J].Journal of Hydrology,2021,596:126082.
[16] STAVROPOULOU M,EXADAKTYLOS G,SARATSIS G.A combined three-dimensional geological-geostatistical-numerical model of underground excavations in rock[J].Rock mechanics and rock engineering,2007,40(3):213-243.
[17] BAO Y,CHEN J,SU L,et al.A novel numerical approach for rock slide blocking river based on the CEFDEM model:A case study from the Samaoding paleolandslide blocking river event[J].Engineering Geology,2023,312:106949.
[18] 郝爱兵，赵伟，郑跃军，等.水文地质调查技术方法发展与应用综述[J].测绘科学，2022,47(8):25-35.HAO A B,ZHAO W,ZHENG Y J,et al.A review on development and application of hydrogeological survey techniques and methods[J].Science of Surveying and Mapping,2022,47(8):25-35.
[19] GROSSMANN T G,KOMOROWSKA U J,LATZ J,et al.Can Physics-Informed Neural Networks beat the finite element method?[J].IMA Journal of Applied Mathematics,2024,81(1):143-174.
[20] ZHANG J,ZHU Y,ZHANG X,et al.Developing a Long Short-Term Memory (LSTM) based model for predicting water table depth in agricultural areas[J].Journal of Hydrology,2018,561:918-929.
[21] PARK C,CHUNG M.Evaluating the groundwater prediction using LSTM model[J].Journal of Korea Water Resources Association,2020,53(4):273-283.
[22] 曹鑫宇，朱琳，宫辉力，等.AM-LSTM网络的北京平原东部地面沉降模拟[J].遥感学报，2022,26(7):1302-1314.CAO X Y,ZHU L,GONG H L,et al.Land subsidence simulation in the east of Beijing plain based on the AM-LSTM Network[J].National Remote Sensing Bulletin,2022,26(7):1302-1314.
[23] YANG J,GRIFFITHS J,ZAMMIT C.National classification of surface-groundwater interaction using random forest machine learning technique[J].River Research and Applications,2019,35(7):932-943.
[24] 孙浩伟，程茹樊，李仲龙，等.基于机器学习模型的南宁盆地地下水中硝酸盐浓度预测及健康风险评价[J].安全与环境工程，2025,32(1):273-280.SUN H W,CHENG R F,LI Z L,et al.Prediction and health risk assessment of groundwater nitrate concentration in Nanning Basin using machine learning models[J].Safety and Environmental Engineering,2025,32(1):273-280.
[25] SAMEEN I,PRADHAN B,LEE S.Self-learning random forests model for mapping groundwater yield in data-scarce areas[J].Natural Resources Research,2019,28:757-775.
[26] LIANG W J,ZHUANG D F,JIANG D,et al.Assessment of debris flow hazards using a Bayesian Network[J].Geomorphology,2012,171:94-100.
[27] LI H,WAN B,CHU D,et al.Progressive geological modeling and uncertainty analysis using machine learning[J].ISPRS International Journal of Geo-Information,2023,12(3):97.
[28] 梁昌玉，吕大炜，初航，等.浅谈中国工程地质学发展现状及未来发展趋势：基于国家自然科学基金(1987—2021)立项数据[J].工程地质学报，2024,32(1):349-357.LIANG C Y,LYU D W,CHU H,et al.Brief discussion on the development status and future trend of engineering geology in China:Based on the project data of national natural science foundation of China[J].Journal of Engineering Geology,2024,32(1):349-357.
[29] 罗袆沅，许强，蒋亚楠，等.基于时序InSAR与机器学习的大范围地面沉降预测方法[J].地球科学，2024,49(5):1736-1745.LUO H Y,XU Q,JIANG Y N,et al.The prediction method of large-scale land subsidence based on multi-temporal InSAR and machine learning[J].Earth Science,2024,49(5):1736-1745.
[30] CHEVALLIER F,CHéRUY F,SCOTT N,et al.A neural network approach for a fast and accurate computation of a longwave radiative budget[J].Journal of Applied Meteorology and Climatology,1998,37(11):1385-1397.
[31] YANG T,SUN F,GENTINE P,et al.Evaluation and machine learning improvement of global hydrological model-based flood simulations[J].Environmental Research Letters,2019,14(11):114027.
[32] KRAFT B,JUNG M,K?RNER M,et al.Towards hybrid modeling of the global hydrological cycle[J].Hydrology and Earth System Sciences Discussions,2021,26(6):1579-1614.
[33] MEKONNEN B A,NAZEMI A,MAZUREK K A,et al.Hybrid modelling approach to prairie hydrology:Fusing data-driven and process-based hydrological models[J].Hydrological Sciences Journal,2015,60(9):1473-1489.
[34] CHO K,KIM Y.Improving streamflow prediction in the WRF-Hydro model with LSTM networks[J].Journal of Hydrology,2022,605:127297.
[35] HUNTER J M,MAIER H R,GIBBS M S,et al.Framework for developing hybrid process-driven,artificial neural network and regression models for salinity prediction in river systems[J].Hydrology and Earth System Sciences,2018,22(5):2987-3006.
[36] ZHU L,LI S,GONG H L,et al.A hybrid numerical method incorporating machine learning into groundwater level model for improving simulation accuracy[J].ACS ES&T Water,2025,5(4):1916-1929.
[37] HE M,JIANG S,REN L,et al.Exploring the performance and interpretability of hybrid hydrologic model coupling physical mechanisms and deep learning[J].Journal of Hydrology,2025,649:132440.
[38] QU P,ZHANG L,ZHU Q,et al.Probabilistic reliability assessment of twin tunnels considering fluid-solid coupling with physics-guided machine learning[J].Reliability Engineering & System Safety,2023,231:109028.
[39] LAI V Q,NGUYEN T K,SHIAU J,et al.Coupling FEA with XGBoost model for estimating uplift resistance of circular anchor in NGI-ADP soils[J].Geotechnical and Geological Engineering,2024,42(1):767-781.
[40] LIU L,ZHOU F,CAO W.Integrate physics-driven dynamics simulation with data-driven machine learning to predict potential targets in maturely explored orefields:A case study in Tongguangshan orefield,Tongling,China[J].Journal of Geochemical Exploration,2024,262:107478.
[41] LIU F,LIU W,LI A,et al.Geotechnical risk modeling using an explainable transfer learning model incorporating physical guidance[J].Engineering Applications of Artificial Intelligence,2024,137(A):109127.
[42] GUAN Q,YANG Z,GUO N,et al.Finite element geotechnical analysis incorporating deep learning-based soil model[J].Computers and Geotechnics,2023,154:105120.
[43] SAMNIOTI A,ANASTASIADOU V,GAGANIS V.Application of machine learning to accelerate gas condensate reservoir simulation[J].Clean Technologies,2022,4(1):153-173.
[44] SHEN H F,JIANG M H,LI J,et al.Coupling model-and data-driven methods for remote sensing image restoration and fusion:Improving physical interpretability[J].IEEE Geoscience and Remote Sensing Magazine,2022,10(2):231-249.
[45] RAISSI M,PERDIKARIS P,KARNIADAKIS G.Physics informed deep learning (part II):Data-driven discovery of nonlinear partial differential equations[EB/OL].https://doi.org/10.48550/arXiv.1711.10566.
[46] RAISSI M,PERDIKARIS P,KARNIADAKIS G E.Physics informed deep learning (part I):Data-driven solutions of nonlinear partial differential equations[EB/OL].https://doi.org/10.48550/arXiv.1711.10561.
[47] RAISSI M,PERDIKARIS P,KARNIADAKIS G E.Physics-informed neural networks:A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations[J].Journal of Computational Physics,2019,378:686-707.
[48] 舒伟，孟胤全，邓芳，等.基于PINNs算法的一维潜水流方程的渗流参数反演[J].南京大学学报(自然科学),2024,60(2):317-327.SHU W,MENG Y Q,DENG F.et al.Inversion of seepage parameters for one-dimensional unconfined aquifer flow equations based on PINNs algorithm[J].Journal of Nanjing University(Natural Science),2024,60(2):317-327.
[49] CUOMO S,COLA S,GIAMPAOLO F,et al.Scientific machine learning through physics-informed neural networks:Where we are and what’s next[J].Journal of Scientific Computing,2022,92(3):88.
[50] KARNIADAKIS G E,KEVREKIDIS I G,LU L,et al.Physics-informed machine learning[J].Nature Reviews Physics,2021,3(6):422-440.
[51] 李野，陈松灿.基于物理信息的神经网络：最新进展与展望[J].计算机科学，2022,49(4):254-262.LI Y,CHEN S C.Physics-informed Neural Networks:Recent advances and prospects[J].Computer Science,2022,49(4):254-262.
[52] HE Q Z,BARAJAS-SOLANO D,TARTAKOVSKY G,et al.Physics-informed neural networks for multiphysics data assimilation with application to subsurface transport[J].Advances in Water Resources,2020,141:103610.
[53] MAMUD M,MUDUNURU M,KARRA S,et al.Quantifying local and global mass balance errors in physics-informed neural networks[J].Scientific Reports,2024,14(1):15541.
[54] ZHANG X,ZHU Y,WANG J,et al.GW-PINN:A deep learning algorithm for solving groundwater flow equations[J].Advances in Water Resources,2022,165:104243.
[55] HARUZI P,MORENO Z.Modeling water flow and solute transport in unsaturated soils using physics-informed neural networks trained with geoelectrical data[J].Water Resources Research,2023,59(6):e2023WR034538.
[56] CUOMO S,ROSA M,Giampaolo F,et al.Solving groundwater flow equation using physics-informed neural networks[J].Computers & Mathematics with Applications,2023,145:106-123.
[57] CHAKRABORTY A,RABINOVICH A,MORENO Z.Physics-informed neural networks for modeling two-phase steady state flow with capillary heterogeneity at varying flow conditions[J].Advances in Water Resources,2024,185:104639.
[58] LAN P,SU J,ZHANG S.Surrogate modeling for unsaturated infiltration via the physics and equality-constrained artificial neural networks[J].Journal of Rock Mechanics and Geotechnical Engineering,2024,16(6):2282-2295.
[59] HE Q,TARTAKOVSKY A.Physics-informed neural network method for forward and backward advection-dispersion equations[J].Water Resources Research,2021,57(7):e2020WR029479.
[60] KANG Q,ZHANG B,CAO Y,et al.Causal prior-embedded physics-informed neural networks and a case study on metformin transport in porous media[J].Water Research,2024,261:121985.
[61] SOULEYMANE Z,LAMIEN B,BEIDARI M,et al.Numerical simulation of pollutants transport in groundwater using deep neural networks informed by physics[J].Gulf Journal of Mathematics,2024,16(2):337-352.
[62] PRADITIA T,KARLBAUER M,OTTE S,et al.Learning groundwater contaminant diffusion-sorption processes with a finite volume neural network[J].Water Resources Research,2022,58(12):e2022WR033149.
[63] SECCI D,GODOY V A,GóMEZ-HERNáNDEZ J J.Physics-Informed Neural Networks for solving transient unconfined groundwater flow[J].Computers & Geosciences,2024,182:105494.
[64] DEPINA I,JAIN S,MAR VALSSON S,et al.Application of physics-informed neural networks to inverse problems in unsaturated groundwater flow[J].Georisk:Assessment and Management of Risk for Engineered Systems and Geohazards,2022,16(1):21-36.
[65] CHEW A W Z,HE R,ZHANG L.Modelling clogging dynamics in groundwater systems using multiscale homogenized physics informed neural network (MHPINN)[J].Engineering Science and Technology,an International Journal,2024,49:101605.
[66] JI Y,ZHA Y,YEH T C J,et al.Groundwater inverse modeling:Physics-Informed Neural Network with disentangled constraints and errors[J].Journal of Hydrology,2024,640:131703.
[67] CHEN Y,XU Y,WANG L,et al.Modeling water flow in unsaturated soils through Physics-Informed Neural Network with principled loss function[J].Computers and Geotechnics,2023,161:105546.
[68] 张升，兰鹏，苏晶晶，等.基于PINNs算法的地下水渗流模型求解及参数反演[J].岩土工程学报，2023,45(2):376-383.ZHANG S,LAN P,SU J J,et al.Simulation and parameter identification of groundwater flow model based on PINNs algorithms[J].Chinese Journal of Geotechnical Engineering,2023,45(2):376-383.
[69] COLA V,BAUDUIN V,BERARDI M,et al.Investigating neural networks with groundwater flow equation loss[J].Mathematics and Computers in Simulation,2025,230:80-93.
[70] WANG N,KONG X Z,ZHANG D.Physics-Informed Convolutional Decoder (PICD):A novel approach for direct inversion of heterogeneous subsurface flow[J].Geophysical Research Letters,2024,51(13):e2024GL108163.
[71] ADOMBI A,CHESNAUX R,BOUCHER M,et al.A causal physics-informed deep learning formulation for groundwater flow modeling and climate change effect analysis[J].Journal of Hydrology,2024,637:131370.
[72] ADOMBI A,CHESNAUX R.Soft causal constraints in groundwater machine learning:A new way to balance accuracy and physical consistency[J].Environmental Earth Sciences,2025,84(2):1-17.
[73] NAGY M,ROTH V.Physics-informed boundary integral networks (PIBI-Nets):A data-driven approach for solving partial differential equations[J].Journal of Computational Science,2024,81:102355.
[74] GUO H,ZHUANG X,RABCZUK T.Stochastic analysis of heterogeneous porous material with modified neural architecture search (NAS) based physics-informed neural networks using transfer learning[EB/OL].https://doi.org/10.48550/arXiv.2010.12344.
[75] YANG Y,MEI G.A deep learning-based approach for a numerical investigation of soil-water vertical infiltration with physics-informed neural networks[J].Mathematics,2022,10(16):2945.
[76] GIAO P.Calculation of pore pressure dissipated from a clay layer due to groundwater extraction using Physics-Informed Neural Network (PINN) with reference to land subsidence analysis[C]//HIEN D,PHUNG D L.International Conference on Geotechnics for Sustainable Infrastructure Development.Singapore:Springer Nature Singapore,2023:2229-2236.
[77] GONG W,ZUO L,LI L,et al.Prediction of stratified ground consolidation via a physics-informed neural network utilizing short-term excess pore water pressure monitoring data[J].Computer-Aided Civil and Infrastructure Engineering,2024,40(2):147-165.
[78] LU Y,MEI G.A deep learning approach for predicting two-dimensional soil consolidation using physics-informed neural networks (PINN)[J].Mathematics,2022,10(16):2949.
[79] 张子龙，潘秋景，仉文岗，等.基于物理信息神经网络的盾构隧道诱发地表沉降预测[J].工程力学，2024,41(4):161-173.ZHANG Z L,PAN Q J,ZHANG W G,et al.Prediction of surface settlements induced by shield tunnelling using physics-informed neural network[J].Engineering Mechanics,2024,41(4):161-173.
[80] ZHENG J,XIE K,WEN C,et al.P-wave prediction method under multi-source spatiotemporal feature fusion and physics-informed neural network[J].Geoenergy Science and Engineering,2023,223:211515.
[81] 李玉伟，李子健，邵力飞，等.基于物理信息约束的页岩油储层可压性评价新方法[J].煤田地质与勘探，2023,51(10):37-51.LI Y W,LI Z J,SHAO L F,et al.A new physics-informed method for the fracability evaluation of shale oil reservoirs[J].Coal Geology & Exploration,2023,51(10):37-51.
[82] ELMORSY M,EL-DAKHAKHNI W,ZHAO B.Rapid permeability upscaling of digital porous media via Physics-Informed Neural Networks[J].Water Resources Research,2023,59(12):e2023WR035064.
[83] 李林，左林龙，胡涛涛，等.基于孔压静力触探试验测试数据的原位固结系数物理信息神经网络反演方法[J].岩土力学，2024,45(10):2889-2899.LI L,ZUO L L,HU T T,et al.A physics-informed neural networks inversion method for in-situ consolidation coefficient based on piezocone penetration test pore pressure data[J].Rock and Soil Mechanics,2024,45(10):2889-2899.
[84] 霍海峰，黄昊宇，李其昂，等.基于PINNs的高度非线性Richards入渗模型研究[J].中国民航大学学报，2023,41(5):6-12.HUO H F,HUANG H Y,LI Q A,et al.Research on highly nonlinear Richards infiltration model based on PINNs[J].Journal of Civil Aviation University of China,2023,41(5):6-12.
[85] YANG L,MENG X,KARNIADAKIS G E.B-PINNs:Bayesian physics-informed neural networks for forward and inverse PDE problems with noisy data[J].Journal of Computational Physics,2021,425:109913.
[86] ZOU Z,MENG X,KARNIADAKIS G E.Correcting model misspecification in physics-informed neural networks (PINNs)[J].Journal of Computational Physics,2024,505:112918.
[87] MOOS J,HANSEL K,ABDULSAMAD H,et al.Robust reinforcement learning:A review of foundations and recent advances[J].Machine Learning and Knowledge Extraction,2022,4(1):276-315.
[88] POURSAEED O,JIANG T,YANG H,et al.Robustness and generalization via generative adversarial training[EB/OL].https://doi.org/10.48550/arXiv.2109.02765.
[89] 潘小果，王凯，邓维鑫.基于NTK理论和改进时间因果的物理信息神经网络加速收敛算法[J].力学学报，2024,56(7):1943-1958.PAN X G,WANG K,DENG W X.Accelerating convergence algorithm for physics-informed neural networks based on NTK theory and modified causality[J].Chinese Journal of Theoretical and Applied Mechanics,2024,56(7):1943-1958.
基本信息:
DOI：10.13928/j.cnki.wrahe.2025.07.002
中图分类号:P642;P641
引用信息:
[1]朱琳,钱陈之皓,宫辉力,等.物理信息神经网络在水文地质与工程地质中的应用研究综述[J].水利水电技术(中英文),2025,56(07):13-25.DOI:10.13928/j.cnki.wrahe.2025.07.002.
基金信息:
国家自然科学基金项目(42271082,42371081)