山西坪上泉水资源量评价及基于SVM理论的泉流量预测模型研究
|
摘要
坪上应急引水工程是山西六大水利工程之组成部分,是山西兴水战略的重要组成部分。坪上泉岩溶水是坪上应急水工程的引水水源。由于该项目的供水目标为忻府、定襄和原平,保证率要求很高。供水工程水源地的建立直接关系着供水区国民经济的可持续发展,以及百万人民的安全用水,因此,山西坪上泉水资源量评价及基于SVM理论的岩溶泉流量预测模型研究具有重要的现实意义和长远的战略意义。
本文在前人研究的基础上,通过对研究区水文、气象、地质、地貌和水文地质条件分析,根据系统理论划分了泉域岩溶水子系统,即甲子湾泉子系统、李家庄泉—坪上村泉子系统、水泉湾泉—段家庄泉子系统和水头沟泉子系统。建立了泉域水文地质概念模型,将泉流量的主要影响因素概化为大气降水、河水渗漏补给和人工开采。
本文简述了支持向量机基本原理,针对支持向量机(SVM)预报模型的智能化程度高、人工干预少,可随变量实测资料的增加,通过自学习来自动调整参数,使模型不断贴近实际,以及操作简单易于工程管理部门实际应用的特点,结合评上泉实际,建立了基于SVM理论的坪上岩溶泉流量预报模型,并针对影响因子对泉水流量时序性作用特征,优选了最佳时间序列长度,确定了最佳核函数,制定了模型预报流程。因此该项研究为引水工程未来数字化管理,为引水工程可供水量预报和供水区的多水源优化配置奠定了基础。
本文还依据坪上泉域地处山区的特点,确定泉水资源量计算方法,即以排泄法为主、补给法校核。计算得:泉域1956～2006系列多年平均岩溶地下水资源量为15200万m3/a;泉域岩溶地下水可利用资源量采用了理论频率分析法,计算结果表明:坪上泉岩溶地下水年可利用资源量为7064万m3/a,地下水资源可采系数为0.465,岩溶水现用量仅占可利用量的4.3%,为有潜力开发区。
?Pingshang jury diversion works is component part of the six large irrigation works in Shanxi province.It is important part of strategy plan for build water conservancy project in Shanxi province.Pingshang karst water is water souce of Pingshang jury diversion works.This project will supply water to Xinfu,Dingxiang and Yuanping.The higher reliability of this project is needed.The construction of the water source is related to sustainable development of national economy in this area and safety water for million people.Therefor the assessment of karst water resource of Pingshang spring and the research of forecast model of karst spring discharge based on SVM theory are very important in practical living and economy development policy.
In this paper based on research results of predecessors the huge data about hydrology,meteorology,geology,geomorphology and hydrogeology are analysed.The karst water system in this area are divided into a certain numbers of subsystems according to the system theory.That is subsystem of Jiaziwan spring,subsystem of Lijiazhuang spring—Pingshangcun spring,subsystem of Shuiquanwan spring—Duanjiazhuang spring,and subsystem of Shuitougou spring.The conceptual hydrogeology model is established.The main effective factors on karst discharge are summarized as rainfall,replenishment by river leakage and human exploitation.
The primary theory of SVM is explained briefly.Because of the SVM forecast model has high intelligence,less human participation,with the observation data increasing the model will modify the parameters automatically,so that to approach the real case,it is easier for operation and easier to use for the department of engineering management,so the discharge model of Pingshang karst spring is established by SVM theory.According to effective factors on spring discharge in time oder,the optimum time oder length and the optimum kernel function are decided.And the process of forecast is decided.The base for diversion works management digitizing,forecast of quantity of water feed and optimization of desposition scheme of water resources in this area is established by this research results.
Because of the Pingshang spring located at mountain area so the computing method for the water quantity evaluation are carefully considered.The drainage method is main method.And the evaluated results are checked by replenishment method.The evaluated results show that the mean value of the underground karst water during 1956 to 2006 is 152 million m3/a.The theoretical frequency analysis is used to evaluate the useable quantity of underground karst water in this spring area.The evaluated results show that the useable quantity of underground karst water in Pingshang is 7.046 million m3/a per year.The useable factor is 0.465.The current consumed water quantity is 4.3 percent only.So this spring area has great potentialities for exploitage.
本文在前人研究的基础上,通过对研究区水文、气象、地质、地貌和水文地质条件分析,根据系统理论划分了泉域岩溶水子系统,即甲子湾泉子系统、李家庄泉—坪上村泉子系统、水泉湾泉—段家庄泉子系统和水头沟泉子系统。建立了泉域水文地质概念模型,将泉流量的主要影响因素概化为大气降水、河水渗漏补给和人工开采。
本文简述了支持向量机基本原理,针对支持向量机(SVM)预报模型的智能化程度高、人工干预少,可随变量实测资料的增加,通过自学习来自动调整参数,使模型不断贴近实际,以及操作简单易于工程管理部门实际应用的特点,结合评上泉实际,建立了基于SVM理论的坪上岩溶泉流量预报模型,并针对影响因子对泉水流量时序性作用特征,优选了最佳时间序列长度,确定了最佳核函数,制定了模型预报流程。因此该项研究为引水工程未来数字化管理,为引水工程可供水量预报和供水区的多水源优化配置奠定了基础。
本文还依据坪上泉域地处山区的特点,确定泉水资源量计算方法,即以排泄法为主、补给法校核。计算得:泉域1956～2006系列多年平均岩溶地下水资源量为15200万m3/a;泉域岩溶地下水可利用资源量采用了理论频率分析法,计算结果表明:坪上泉岩溶地下水年可利用资源量为7064万m3/a,地下水资源可采系数为0.465,岩溶水现用量仅占可利用量的4.3%,为有潜力开发区。
?Pingshang jury diversion works is component part of the six large irrigation works in Shanxi province.It is important part of strategy plan for build water conservancy project in Shanxi province.Pingshang karst water is water souce of Pingshang jury diversion works.This project will supply water to Xinfu,Dingxiang and Yuanping.The higher reliability of this project is needed.The construction of the water source is related to sustainable development of national economy in this area and safety water for million people.Therefor the assessment of karst water resource of Pingshang spring and the research of forecast model of karst spring discharge based on SVM theory are very important in practical living and economy development policy.
In this paper based on research results of predecessors the huge data about hydrology,meteorology,geology,geomorphology and hydrogeology are analysed.The karst water system in this area are divided into a certain numbers of subsystems according to the system theory.That is subsystem of Jiaziwan spring,subsystem of Lijiazhuang spring—Pingshangcun spring,subsystem of Shuiquanwan spring—Duanjiazhuang spring,and subsystem of Shuitougou spring.The conceptual hydrogeology model is established.The main effective factors on karst discharge are summarized as rainfall,replenishment by river leakage and human exploitation.
The primary theory of SVM is explained briefly.Because of the SVM forecast model has high intelligence,less human participation,with the observation data increasing the model will modify the parameters automatically,so that to approach the real case,it is easier for operation and easier to use for the department of engineering management,so the discharge model of Pingshang karst spring is established by SVM theory.According to effective factors on spring discharge in time oder,the optimum time oder length and the optimum kernel function are decided.And the process of forecast is decided.The base for diversion works management digitizing,forecast of quantity of water feed and optimization of desposition scheme of water resources in this area is established by this research results.
Because of the Pingshang spring located at mountain area so the computing method for the water quantity evaluation are carefully considered.The drainage method is main method.And the evaluated results are checked by replenishment method.The evaluated results show that the mean value of the underground karst water during 1956 to 2006 is 152 million m3/a.The theoretical frequency analysis is used to evaluate the useable quantity of underground karst water in this spring area.The evaluated results show that the useable quantity of underground karst water in Pingshang is 7.046 million m3/a per year.The useable factor is 0.465.The current consumed water quantity is 4.3 percent only.So this spring area has great potentialities for exploitage.
引文
[1]《河南省水资源综合评价工作大纲》,2001年10月,河南省水利厅编制
[2]刘克岩,徐斌,米玉华.水资源水质水量结合评价方法及其应用.河北水利水电技术,2003,(3):6-8
[3]刘克岩,王桂玲.以用水为主体的水质水量结合水资源评价方法.水文,2003,22(3):32-33
[4]梁德华,蒋火华.河流水质综合评价方法的统一和改进。中国环境检测,2002,18(2):63-66
[5]蒋火华,朱建平,梁德华等.综合污染指数评价与水质类别判定的关系.中国环境检测,1999,15(6):46-48
[6]岩溶水系统——山西岩溶大泉研究,中国地质科学院岩溶地质研究所、山西省水资源管理委员会办公室,1993
[7]山西省泉域边界范围及重点保护区,山西省水资源管理委员会办公室,1998
[8]山西省南庄引水工程坪上泉岩溶水系统勘查及资源评价研究报告,山西省水利厅南庄引水工程指挥部,山西省忻州地区水利勘察设计院,山西省地质局工程勘察施工公司,1995
[9]山西省忻州市水资源评价,山西省忻州市水资源管理委员会,山西省忻州市水局,2004
[10]山西省忻州市地表水资源评价,山西省忻州市水资源管理委员会,山西省忻州市水局,2004
[11]山西省忻州市地下水资源调查评价报告,山西省忻州市水资源管理委员会,山西省忻州市水局,2004
[12]滹沱河西龙池抽水蓄能电站可行性研究前期站址选择工程地质勘察报告,能源部北京勘察设计院,山西省水利勘察设计院,1994
[13]忻州地区1/20万水文地质图,山西省地质调查矿局水文地质队,1974
[14]山西省滹沱河南庄引水工程环境影响评价报告,山西省环境保护研究所,1991
[15]盂县幅”、“平型关幅”区域地质调查报告,山西省地矿局区域地质调查队,1965
[16]Vapnik V.The nature of statistical learning theory[M].New York:Springer—Verlag,1995.
[17]Cortes C,Vapnik V.Support vector networks[J].Machine learning,1995,20(1):273-297.
[18]刘向东,陈兆乾.基于支持向量机方法的人脸识别研究[J].小型微型计算机系统,2004,25(12):2 261- 2 263.
[19]张爱丽,刘广利,刘长宇.基于SVM的多类文本分类研究[J].情报杂志,2004.23(9):6-10.
[20]魏新,冯兴杰,刘山.基于支持向量机的多元文本分类研究[J].海军工程大学学报.2004,16(5):30-32.
[21]Smola A,Scholkopf B.On a kernel based method for pattern recognition,regression,approximation and operator inversion[J].Algorithmica,1998,22(1):211- 231.
[22] Gill P E,Murray W ,Wright M H.Practical optimization[M].New York: Academic Press,1981.
[23]Platt J.Fast Training of Support Vector Machines Using Sequential Minimal Optimization[A].Scholkopf B,Burges C,Smola A.Ad—vances in Kernel Methods—Support Vector Learning[C].Cambridge MA:MIT Press, 1999.
[24]Fei S,Lawrence K,Daniel D.Multiplicative updates for nonnegative quadratic programming in support vector machines[A].suztlnnaB, Sebastian T,Klaus O.Advances in Neural Information Processing Systems[C].Cambridge MA:MIT Press,2003.
[25]Jinbo B,Bennett K-Emhrechts M.Dimensionality Reduction via Sparse Support Vector Machines[J].J0urnal of Machine LearningResearch,2003。3(1):1 229—1 243.
[26]Lin C J.Asymptotic convergence of an SMO algorithm without any assumptions[J].IEEE Transactions on Neural Netw0rks.2001.13(1):248—250.
[27]李建民,张钹林,福宗.支持向量机的训练算法[J].清华大学学报(自然科学版),2003,43(1):12O-124.
[28]Bennett K,Campbell C.Support vector machineshype0rhallelujah[J].SIGKDD Explorations,2000,2(2);1—13.
[29]Jinbo B.Kristin B.Duality,Geometry and Support Vector Regression[A].Dietterich T,Becker S,Ghahramaniz.Advances in Neural Information Processing Systems[C].Cambridge MA:MIT Press,2001.
[30]Keerthi S,Shevade S,Bhattcharyya C.A fast iterative nearest point algorithm for support vector machine classifier design[J].IEEE Transactions on Neural Network,2000,l1(1):124—136.
[31] Erin B.Duality and Geometry in SVM Classifiers[A].Pat L.Proceedings of the Seventeenth International Conference on MachineLearning[C].San Francisco:Morgan Kaufmann Press,2000.
[32]张铃.支持向量机理论与基于规划的神经网络学习算法[J].计算机学报2001,24(2):l13—118.
[33]Evgeniou T,Pontil M,Poggio T.Regularization networks and support vector machines[J].Advances in Computational Mathematics,2000,13(1): 1—5O.
[2]刘克岩,徐斌,米玉华.水资源水质水量结合评价方法及其应用.河北水利水电技术,2003,(3):6-8
[3]刘克岩,王桂玲.以用水为主体的水质水量结合水资源评价方法.水文,2003,22(3):32-33
[4]梁德华,蒋火华.河流水质综合评价方法的统一和改进。中国环境检测,2002,18(2):63-66
[5]蒋火华,朱建平,梁德华等.综合污染指数评价与水质类别判定的关系.中国环境检测,1999,15(6):46-48
[6]岩溶水系统——山西岩溶大泉研究,中国地质科学院岩溶地质研究所、山西省水资源管理委员会办公室,1993
[7]山西省泉域边界范围及重点保护区,山西省水资源管理委员会办公室,1998
[8]山西省南庄引水工程坪上泉岩溶水系统勘查及资源评价研究报告,山西省水利厅南庄引水工程指挥部,山西省忻州地区水利勘察设计院,山西省地质局工程勘察施工公司,1995
[9]山西省忻州市水资源评价,山西省忻州市水资源管理委员会,山西省忻州市水局,2004
[10]山西省忻州市地表水资源评价,山西省忻州市水资源管理委员会,山西省忻州市水局,2004
[11]山西省忻州市地下水资源调查评价报告,山西省忻州市水资源管理委员会,山西省忻州市水局,2004
[12]滹沱河西龙池抽水蓄能电站可行性研究前期站址选择工程地质勘察报告,能源部北京勘察设计院,山西省水利勘察设计院,1994
[13]忻州地区1/20万水文地质图,山西省地质调查矿局水文地质队,1974
[14]山西省滹沱河南庄引水工程环境影响评价报告,山西省环境保护研究所,1991
[15]盂县幅”、“平型关幅”区域地质调查报告,山西省地矿局区域地质调查队,1965
[16]Vapnik V.The nature of statistical learning theory[M].New York:Springer—Verlag,1995.
[17]Cortes C,Vapnik V.Support vector networks[J].Machine learning,1995,20(1):273-297.
[18]刘向东,陈兆乾.基于支持向量机方法的人脸识别研究[J].小型微型计算机系统,2004,25(12):2 261- 2 263.
[19]张爱丽,刘广利,刘长宇.基于SVM的多类文本分类研究[J].情报杂志,2004.23(9):6-10.
[20]魏新,冯兴杰,刘山.基于支持向量机的多元文本分类研究[J].海军工程大学学报.2004,16(5):30-32.
[21]Smola A,Scholkopf B.On a kernel based method for pattern recognition,regression,approximation and operator inversion[J].Algorithmica,1998,22(1):211- 231.
[22] Gill P E,Murray W ,Wright M H.Practical optimization[M].New York: Academic Press,1981.
[23]Platt J.Fast Training of Support Vector Machines Using Sequential Minimal Optimization[A].Scholkopf B,Burges C,Smola A.Ad—vances in Kernel Methods—Support Vector Learning[C].Cambridge MA:MIT Press, 1999.
[24]Fei S,Lawrence K,Daniel D.Multiplicative updates for nonnegative quadratic programming in support vector machines[A].suztlnnaB, Sebastian T,Klaus O.Advances in Neural Information Processing Systems[C].Cambridge MA:MIT Press,2003.
[25]Jinbo B,Bennett K-Emhrechts M.Dimensionality Reduction via Sparse Support Vector Machines[J].J0urnal of Machine LearningResearch,2003。3(1):1 229—1 243.
[26]Lin C J.Asymptotic convergence of an SMO algorithm without any assumptions[J].IEEE Transactions on Neural Netw0rks.2001.13(1):248—250.
[27]李建民,张钹林,福宗.支持向量机的训练算法[J].清华大学学报(自然科学版),2003,43(1):12O-124.
[28]Bennett K,Campbell C.Support vector machineshype0rhallelujah[J].SIGKDD Explorations,2000,2(2);1—13.
[29]Jinbo B.Kristin B.Duality,Geometry and Support Vector Regression[A].Dietterich T,Becker S,Ghahramaniz.Advances in Neural Information Processing Systems[C].Cambridge MA:MIT Press,2001.
[30]Keerthi S,Shevade S,Bhattcharyya C.A fast iterative nearest point algorithm for support vector machine classifier design[J].IEEE Transactions on Neural Network,2000,l1(1):124—136.
[31] Erin B.Duality and Geometry in SVM Classifiers[A].Pat L.Proceedings of the Seventeenth International Conference on MachineLearning[C].San Francisco:Morgan Kaufmann Press,2000.
[32]张铃.支持向量机理论与基于规划的神经网络学习算法[J].计算机学报2001,24(2):l13—118.
[33]Evgeniou T,Pontil M,Poggio T.Regularization networks and support vector machines[J].Advances in Computational Mathematics,2000,13(1): 1—5O.