基于KELM地面沉降替代模型的地下水多目标管理模型研究
|
摘要
基于核函数极限学习机(Kernelextreme learning machine,KELM)方法建立地下水-地面沉降耦合的替代模型,与混合多目标算法(Non-dominated Sorting Genetic Algorithm II,NSGAII)相耦合,实现在地面沉降约束下(沉降速率和地下水位红线)地下水资源合理开发利用和地面沉降防控减灾的多目标优化.以三维均质多层含水系统中抽水引发的地面沉降算例为对象,采用MODFLOW-2005中的地面沉降模拟子程序(Subsidence for the water table,SUB-WT)模拟地面沉降过程,基于KELM方法,采用线下和线上两种模式建立替代模型,分别构建了基于线下地面沉降替代模型的多目标管理模型(KELM model based multi-objective optimization model for land subsidence management,KELM&MOLS)和基于自适应(线上)替代模型的多目标管理模型(Adaptive KELM&MOLS,AKELM&MOLS).模拟优化结果显示:(1)基于线上模式训练的替代模型的模拟精度更高,拟合相关系数达0.9988以上,基本接近SUB-WT模拟模型的评价精度;(2)KELM&MOLS优化求解效率提高了15倍,但其搜索的Pareto解的质量最差,AKELM&MOLS求解效率提高了3倍,同时保证了优化解的收敛性和精度.
A combined simulation-optimization(S-O) model that integrates the Nondominated Sorting Genetic Algorithm II(NSGAII)with kernel extreme learning machine(KELM)based surrogate model was developed for deriving multiple management strategies for utilization of groundwater resources and disaster mitigation considering the constraints of the rate of land subsidence and groundwater level. In the combined S-O model,the SUB-WT,which is the compaction package of MODFLOW-2005,was utilized to simulate the process of land subsidence induced by groundwater extraction in three-dimensional homogeneous multi-layered aquifer system. The KELM was developed based on off-line and on-line framework and evaluated as an approximate simulator to generate the patterns of groundwater level and land subsidence for reducing huge computational burden. After that,the KELM model based multi-objective optimization model(KELM&MOLS)and the adaptive KELM&MOLS(AKELM&MOLS)for land subsidence management were established and evaluated through a synthetic example application. The simulation results indicate that the adaptive KELM get higher fitting accuracy close to the SUB-WT simulation model,and the correlation coefficient is above 0.9988. The optimization results showed that the AKELM&MOLS did not only improve the prediction accuracy of Pareto-optimal solutions compared with those by the KELM&MOLS,but also maitained the equivalent quality of Pareto optimal solutions compared with those by NSGAII coupled with the original simulation model.
A combined simulation-optimization(S-O) model that integrates the Nondominated Sorting Genetic Algorithm II(NSGAII)with kernel extreme learning machine(KELM)based surrogate model was developed for deriving multiple management strategies for utilization of groundwater resources and disaster mitigation considering the constraints of the rate of land subsidence and groundwater level. In the combined S-O model,the SUB-WT,which is the compaction package of MODFLOW-2005,was utilized to simulate the process of land subsidence induced by groundwater extraction in three-dimensional homogeneous multi-layered aquifer system. The KELM was developed based on off-line and on-line framework and evaluated as an approximate simulator to generate the patterns of groundwater level and land subsidence for reducing huge computational burden. After that,the KELM model based multi-objective optimization model(KELM&MOLS)and the adaptive KELM&MOLS(AKELM&MOLS)for land subsidence management were established and evaluated through a synthetic example application. The simulation results indicate that the adaptive KELM get higher fitting accuracy close to the SUB-WT simulation model,and the correlation coefficient is above 0.9988. The optimization results showed that the AKELM&MOLS did not only improve the prediction accuracy of Pareto-optimal solutions compared with those by the KELM&MOLS,but also maitained the equivalent quality of Pareto optimal solutions compared with those by NSGAII coupled with the original simulation model.
引文
[1] Koster K,Erkens G,Zwanenburg C.A new soil mechanics approach to quantify and predict land subsidence by peat compression.Geophysical Research Letters,2016,43(20):10792-10799.
[2] Guo H P,Zhang Z C,Cheng G M,et al.Groundwater-derived land subsidence in the North China Plain.Environmental Earth Sciences,2015,74(2):1415-1427.
[3] Cao G L,Han D M,Moser J.Groundwater exploitation management under land subsidence constraint:Empirical evidence from the Hangzhou-Jiaxing-Huzhou Plain,China.Environmental Management,2013,51(6):1109-1125.
[4] 闫世龙,王焰新,马腾等.内陆新生代断陷盆地区地面沉降机理及模拟——以山西省太原市为例.武汉:中国地质大学出版,2006,2-3.(Yan S L,Wang Y X,Ma T,et al.Mechanism and simulation of land subsidence in the Cenozoic inland faulted basin:A case study of Taiyuan City,Shanxi Province,China.Wuhan:China University of Geosciences Press,2006,2-3.)
[5] 薛禹群.论地下水超采与地面沉降.地下水,2012,6:1-5.(Xue Y Q.Discussion on groundwater overexploitation and ground settlement.Ground Water,2012,6:1-5.)
[6] Teatini P,Tosi L,Strozzi T.Quantitative evidence that compaction of Holocene sediments drives the present land subsidence of the Po Delta,Italy.Journal of Geophysical Research:Solid Earth,2011,116(B8):B08407,doi:org/10.1029/2010JB008122
[7] Cui Z.Land subsidence induced by the engineering-environmental effect.Springer Berlin Heidelberg,2018,10-12.
[8] 杨蕴,朱琳,林锦等.考虑地面沉降约束的地下水模拟优化管理模型.南京大学学报(自然科学),2016,52(3):470-478.(Yang Y,Zhu L,Lin J,et al.Simulation-optimization modeling for groundwater management considering land subsidence.Journal of Nanjing University(Natural Sciences),2016,52(3):470-478.)
[9] 宋健,吴剑锋,杨蕴等.基于含水层DNAPL污染修复替代模型的多目标优化研究.中国环境科学,36(11):3390-3396.(Song J,Wu J F,Yang Y,et al.A Kriging-based surrogate model for multi-objective optimization of DNAPL-contaminated aquifer remediation.China Environmental Science,2016,36(11):3390-3396.)
[10] Hussain M S,Javadi A A,Ahangar-Asr A,et al.A surrogate model for simulation-optimization of aquifer systems subjected to seawater intrusion.Journal of Hydrology,2015,523:542-554.
[11] Chen C W,Wei C C,Liu H J,et al.Application of neural networks and optimization model in conjunctive use of surface water and groundwater.Water Resources Management,2014,28(10):2813-2832.
[12] Ketabchi H,Ataie-Ashtiani B.Evolutionary algorithms for the optimal management of coastal groundwater:a comparative study toward future challenges.Journal of Hydrology,2015,520:193-213.
[13] Song J,Yang Y,Wu J F,et al.Adaptive surrogate model based multiobjective optimization for coastal aquifer management.Journal of Hydrology,2018,561:98-111.
[14] Leake S A,Galloway D L.MODFLOW groundwater mode-user guide to the subsidence and aquifer-system compaction Package(SUB-WT)for water-table aquifers.U S Geological Survey Techniques and Methods 6-A23,2016-12-02.https://pnbs.usgs.gov/tm/2007/06A23/.
[15] Leake S A,Galloway D L.Use of the SUB-WT package for MODFLOW to simulate aquifer-system compaction in Antelope Valley,California,USA ∥ Carreon-Freyre D,Cerca M,Callongn D L.Land subsidence,associated hazards and the role of natural resources development:Proceedings.Santiago de Querétaro,Mexico:IAHS Publication,2010,39:61-67.
[16] Zhu L,Gong H L,Li X J,et al.Land subsidence due to groundwater withdrawal in the northern Beijing plain,China.Engineering Geology,2015,193:243-255.
[17] 杜思思.海河平原地下水与地面沉降模型模拟研究.博士学位论文.北京:中国地质大学(北京),2011.(Du S S.Study on the model of groundwater and land subsidence in Haihe river basin.Ph.D.Dissertation.Beijing:China University of Geosciences(Beijing),2011.)
[18] Huang G B,Zhu Q Y,Siew C K.Extreme learning machine:theory and applications.Neurocomputing,2006,70(1-3):489-501.
[19] Deb K,Pratap A,Agarwal S,et al.A fast and elitist multiobjective genetic algorithm:NSGA-Ⅱ.IEEE Transactions on Evolutionary Computation,2006,6(2):182-197.
[2] Guo H P,Zhang Z C,Cheng G M,et al.Groundwater-derived land subsidence in the North China Plain.Environmental Earth Sciences,2015,74(2):1415-1427.
[3] Cao G L,Han D M,Moser J.Groundwater exploitation management under land subsidence constraint:Empirical evidence from the Hangzhou-Jiaxing-Huzhou Plain,China.Environmental Management,2013,51(6):1109-1125.
[4] 闫世龙,王焰新,马腾等.内陆新生代断陷盆地区地面沉降机理及模拟——以山西省太原市为例.武汉:中国地质大学出版,2006,2-3.(Yan S L,Wang Y X,Ma T,et al.Mechanism and simulation of land subsidence in the Cenozoic inland faulted basin:A case study of Taiyuan City,Shanxi Province,China.Wuhan:China University of Geosciences Press,2006,2-3.)
[5] 薛禹群.论地下水超采与地面沉降.地下水,2012,6:1-5.(Xue Y Q.Discussion on groundwater overexploitation and ground settlement.Ground Water,2012,6:1-5.)
[6] Teatini P,Tosi L,Strozzi T.Quantitative evidence that compaction of Holocene sediments drives the present land subsidence of the Po Delta,Italy.Journal of Geophysical Research:Solid Earth,2011,116(B8):B08407,doi:org/10.1029/2010JB008122
[7] Cui Z.Land subsidence induced by the engineering-environmental effect.Springer Berlin Heidelberg,2018,10-12.
[8] 杨蕴,朱琳,林锦等.考虑地面沉降约束的地下水模拟优化管理模型.南京大学学报(自然科学),2016,52(3):470-478.(Yang Y,Zhu L,Lin J,et al.Simulation-optimization modeling for groundwater management considering land subsidence.Journal of Nanjing University(Natural Sciences),2016,52(3):470-478.)
[9] 宋健,吴剑锋,杨蕴等.基于含水层DNAPL污染修复替代模型的多目标优化研究.中国环境科学,36(11):3390-3396.(Song J,Wu J F,Yang Y,et al.A Kriging-based surrogate model for multi-objective optimization of DNAPL-contaminated aquifer remediation.China Environmental Science,2016,36(11):3390-3396.)
[10] Hussain M S,Javadi A A,Ahangar-Asr A,et al.A surrogate model for simulation-optimization of aquifer systems subjected to seawater intrusion.Journal of Hydrology,2015,523:542-554.
[11] Chen C W,Wei C C,Liu H J,et al.Application of neural networks and optimization model in conjunctive use of surface water and groundwater.Water Resources Management,2014,28(10):2813-2832.
[12] Ketabchi H,Ataie-Ashtiani B.Evolutionary algorithms for the optimal management of coastal groundwater:a comparative study toward future challenges.Journal of Hydrology,2015,520:193-213.
[13] Song J,Yang Y,Wu J F,et al.Adaptive surrogate model based multiobjective optimization for coastal aquifer management.Journal of Hydrology,2018,561:98-111.
[14] Leake S A,Galloway D L.MODFLOW groundwater mode-user guide to the subsidence and aquifer-system compaction Package(SUB-WT)for water-table aquifers.U S Geological Survey Techniques and Methods 6-A23,2016-12-02.https://pnbs.usgs.gov/tm/2007/06A23/.
[15] Leake S A,Galloway D L.Use of the SUB-WT package for MODFLOW to simulate aquifer-system compaction in Antelope Valley,California,USA ∥ Carreon-Freyre D,Cerca M,Callongn D L.Land subsidence,associated hazards and the role of natural resources development:Proceedings.Santiago de Querétaro,Mexico:IAHS Publication,2010,39:61-67.
[16] Zhu L,Gong H L,Li X J,et al.Land subsidence due to groundwater withdrawal in the northern Beijing plain,China.Engineering Geology,2015,193:243-255.
[17] 杜思思.海河平原地下水与地面沉降模型模拟研究.博士学位论文.北京:中国地质大学(北京),2011.(Du S S.Study on the model of groundwater and land subsidence in Haihe river basin.Ph.D.Dissertation.Beijing:China University of Geosciences(Beijing),2011.)
[18] Huang G B,Zhu Q Y,Siew C K.Extreme learning machine:theory and applications.Neurocomputing,2006,70(1-3):489-501.
[19] Deb K,Pratap A,Agarwal S,et al.A fast and elitist multiobjective genetic algorithm:NSGA-Ⅱ.IEEE Transactions on Evolutionary Computation,2006,6(2):182-197.