摘要
污染源项反演是海域水质模型验证工作的主要难点之一.本文根据污染源项和监测点位的地理位置及空间分布将研究海域划分为若干单元,建立海域组合单元水质模型,并与基于人工神经网络的数据驱动模型有机结合,提出污染源浓度优化反演的新方法:通过多污染源项设计算例的水质模型计算,构建海域内部监测点污染物浓度响应集;以数据驱动模型建立监测点同污染源项之间浓度的非线性关系;将实测资料带入非线性关系中,获得多源项匹配关系,进行污染源浓度的优化反演研究.以渤海湾海域水质模型7个污染源浓度的优化反演为例,采用"孪生"试验验证反演新方法的可行性,结果表明该方法是有效的.
Source inversion of marine water quality model is one of major difficulties for model calibration.A marine water quality model was established with integrated elements,which was divided by the geographical position and spatial distribution of pollution sources and gauge stations.In this paper,a pollution source concentration optimal inversion method was developed by combination of the marine integrated element water quality model and data-driven model based on artificial neural network.Response sets of internal gauge stations were built up by numerical simulation results of multiple pollution sources design cases.The nonlinear relationship of concentration between gauge stations and pollution sources was obtained by data driven model based on artificial neuron network,and filed data was imported into the relationship for inversing pollution sources and the matching relationship in them dynamically.A case study in the Bohai Bay with seven pollution sources and twin experiment was presented for validating the presented method.Verification results show the method is efficient.
引文
[1]Ganoulis J G.Engineering risk analysis of water pollution[M].New York:VCH,1994
[2]Takyi A K,Lence B J.Surface water quality management using a multiple realization chance constrained method[J].Water Resources Research,1999,35(5):1657-1669
[3]Huang C H,Li J X,Kim S.An inverse problem in estimating the strength of contaminant source for groundwater systems[J].Applied Mathematical Modelling,2008,32(4):417-431
[4]Li G S,Liu J Q,Fan X P,et al.A new gradient regularization algorithm for source term inversion in 1D transportationwith final observations[J].Applied Mathematics and Computation,1996,196(2):646-660
[5]韩龙喜.河道一维污染源控制反问题[J].水科学进展,2001,12(1):39-44Han Longxi.Inverse problem on amount of pollutant into natural channel[J].Advances in Water Science,2001,12(1):39-44
[6]闵涛,周孝德,张世梅,等.对流-扩散方程源项识别反问题的遗传算法[J].水动力学研究与进展,2004,19(4):520-524Min Tao,Zhou Xiaode,Zhang Shimei,et al.Genetic algorithm to an inverse problem of source term identification forconvection-diffusion equation[J].Chinese Journal of Hydrodynamics,2004,19(4):520-524
[7]朱嵩,刘国华,王立忠,等.水动力-水质耦合模型污染源项识别的贝叶斯方法[J].四川大学学报(工程科学版),2009,41(5):30-35Zhu Song,Liu Guohua,Wang Lizhong,et al.A baysian approach for the identification of pollution source in waterquality model coupled with hydrodynamics[J].Journal of Sichuan University,2009,41(5):30-35
[8]Solomatine D P.Data-driven modelling:paradigm,methods,experiences[C].Proc.5th International Conference onHydroinformatics,Cardiff,UK,2002:757-763
[9]Rumelhart D E,Hinton G E,Williams R J.Learning representations by back propagating errors[J].Nature,1986,323,533-536
[10]Hornik K,Stinchcombe M,White H.Multilayer feedforward networks are universal approximators[J].Neural Networks,1988,2(5):359-475
[11]李明昌,梁书秀,孙昭晨.异常天气条件下潮位过程神经网络补遗预测方法的研究[J].计算力学学报,2008,25(3):368-372Li Mingchang,Liang Shuxiu,Sun Zhaochen.Research on tide supplemental prediction using ANN methods underunusual weather[J].Chinese Journal of Computational Mechanics,2008,25(3):368-372
[12]Moruto A,Nakastuji K,Huh J Y.A numerical study of three-dimensional buoyant surface jet[J].Proc.6th Congress,APD—IAHR,1988,3:57-64
[13]李明昌,梁书秀,孙昭晨.海域潮汐模型开边界反演优化方法研究[J].水动力学研究与进展,2008,23(6):646-654Li Mingchang,Liang Shuxiu,Sun Zhaochen.Study on optimization inversion method for tidal model with open boundarycondition[J].Chinese Journal of Hydrodynamics,2008,23(6):646-654
[14]赵俊生,王桂云,陈则实,等.海洋环境有效利用的分析研究——北部湾广西沿岸海域[M].北京:海洋出版社,2002Zhao Junsheng,Wang Guiyun,Cheng Zeshi,et al.Analysis research on effective utilization of marine environment[M].Beijing:Ocean Press,2002
[15]徐青,刘玉光,程永存,等.伴随同化技术在渤、黄海生态模型中的应用:控制变量的选取与孪生实验[J].高技术通报,2006,16(1):78-83Xu Qing,Liu Yuguang,Cheng Yongcun,et al.Adjoint assimilation technique in a marine ecosystem model in the BohaiSea and the Yelloe Sea:control variables and twin experiment[J].High Technology Letters,2006,16(1):78-83