基于遗传算法-BP神经网络的水库富营养化研究
|
摘要
当今社会,水质性缺水日渐突出,保护饮用水水源水质迫在眉睫。富营养化己成为一个全球性的重大水环境问题,全球经济持续高速增长的同时,带来了日益严重的环境污染问题。其中出现了湖泊、水库、河流等水体富营养化日益严重的现象,特别是作为城市饮用水源的水库,水体富营养化过程加快,水质恶化,严重危害人们的健康。万宁水库位于海南省万宁市境内,是1959年修建的大型水库,是一宗以饮用水源为主,兼有灌溉、发电、防洪、养鱼等综合利用的大型水利工程。
随着万宁水库整个流域开发活动的加剧,必然带来大量的对其生态环境不利的问题。因此掌握水体富营养化规律,并对其进行准确预测则显得尤为重要。本文在总结和借鉴前人研究成果的基础上,主要就以下几方面进行了研究:
①本文对万宁水库的水质现状进行了调查研究、收集资料等工作。用因子分析法筛选和分析收集到的资料,找出了影响水库富营养化的主要因子:水温、总磷、总氮、溶解氧、叶绿素a;绘制出万宁水库主要水质参数的分布规律曲线。
②从水库富营养化水平短期预测和营养盐长期预测的角度建立了万宁水库富营养化预测系统。
③为了准确、有效的预测预警,进行水库富营养化短期预测,分析现有预测模型的长处与不足,结合前面分析的数据规律选取了最优模型-BP神经网络,用自适应遗传算法对BP神经网络的不足予以修正。
通过因子分析分析找出预测指标-叶绿素a、水温、总磷、总氮、溶解氧作为模型输入,下月叶绿素a作为输出建立自适应GABP神经网络数学模型。
用Matlab7.0编写程序,采用取水点2000-2005年月平均监测数据对模型进行训练,发现模型拟合度、泛化能力较好:神经网络经865次学习后,误差达到预设精度0.0001,运行总时间仅为36.5770s;拟合值与实测值之间的相关关系系数(R)等于0.999。
模型与没经过自适应遗传算法优化的BP神经网络的训练结果比较发现:自适应学习速率BP神经网络经27950次学习后,误差才收敛到0.001857,除了耗费大量的时间,精度仍不满足要求,两者误差相差一个数量级。
采用修正后的2006年月均监测数据对训练后的自适应GABP模型进行实际预测,最不利的相对误差仅为-11.4%,只有一个相对误差绝对值大于10%,平均相对误差0.172%,预测精度较高,可作为水库富营养化水平预测的依据。
利用自适应GABP的短期预测的精度,对水库富营养化水平预测预警,经过分析,警报点选取为叶绿素a的浓度0.004mg/L,并对超过警报点的预测提出应急措施。
④同时还建立了狄龙模型来研究水库营养盐中长期变化趋势,对中长期TN、TP年平均浓度进行预测,发现狄龙模型精度虽然没有富营养化短期预测精度高,但能够满足中长期预测的精度要求;然后利用经典水质模型计算环境容量,在上述中长期预测的基础上得出万宁水库TN、TP的削减计划,最后提出污染防治措施。
Nowadays the water quality is of gradual shortage, the protection of drinking water is imminent. Eutrophication has become a major global water environment, the global economy maintained rapid growth, has brought increasingly serious problem of environmental pollution. The world’s fast economical development has resulted in more and more seriously Environmental pollution problems such as water eutrophication of lakes and reservoir. In the areas of lots of human activities with the eutrophication is speeding up, the water quality is getting so worse and worse that the development of society and economy is limited. Located in Wanning City in Hainan Province, Wanning Reservoir was built in 1959. It provides services of irrigation, the main source of drinking water, power generation, flood control, fish and so on.
As the Wanning reservoir basin-wide development activities intensify, a large number of ecological environment of negative issues will inevitably be brought about. Therefore it is particularly important to master of eutrophication and predict it accurately. In this paper, to sum up and learn research results from their predecessors, the following points were mainly studied:
①The progress and research methods of the eutrophication is systematically expounded.
②from short-term and long-term forecasts predict the perspective of the establishment of the Wanning reservoir eutrophication forecast management system.
③to accurate and effective early-warning forecast for short-term forecasting, prediction analysis of existing strengths and weaknesses of the model, combining the analysis of the data in front of selecting the optimal model - BP neural networks, genetic algorithms using the BP neural network to be less than that .
In this paper, the Wanning reservoir water quality status of the investigations and studies, collection of information are done. Use factor analysis screening and analysis of the information gathered to identify the impact of eutrophication the main factor Reservoir: water temperature, total phosphorus, total nitrogen, dissolved oxygen, chlorophyll a; drawn Wanning Reservoir water quality parameters of the distribution curve, analyzed Eutrophication-made reservoir, analysis of the causes of eutrophication: in Wanning reservoir in recent years nitrogen, phosphorus have been exceeded the standards; serious Wanning reservoir in the tropical oceans of the monsoon climate, the perennial water temperature higher, the light intensity; slow water flow.
Analysis of the strengths of the existing prediction model and the lack of analysis of the data in front of selecting the optimal model - BP neural networks, genetic algorithms using the BP neural network to be less than that.
Through the mechanism of eutrophication forecast to identify indicators - chlorophyll a, water temperature, total phosphorus, total nitrogen, dissolved oxygen, as a model input, next month as output Chl-a neural network to establish adaptive GABP mathematical model.
With Matlab7.0 preparation procedures, use of water years 2000-2005, an average of monitoring data on the model of training for training and found fitting model, the ability to better generalization: neural networks, the 865 study, the error to a default Accuracy of 0.0001, run total time was 36.5770s; fitting and measured the relationship between the coefficient (R) is 0.999.
Adaptive GABP model and BP neural network optimized not by adaptive genetic algorithm training results found: adaptive learning rate BP neural network by 27,950 after learning of the error before converging to 0.001857, in addition to a significant amount of time, not accuracy meet the requirements of error of the difference from an order of magnitude.
After adaptive GABP model training, forecast by the 2006 years of amended monitoring data: the biggest relative error was -11.4%, only a relative error of absolute value greater than 10 percent, the average relative error of 0.172 percent, Forecast for high precision, the reservoir can be used as the basis for the forecast state of eutrophication.
Adaptive GABP use of short-term forecast accuracy, forecast early warning of water quality, after analysis, warning line selected as the concentration of chlorophyll a 0.004 mg / L, exceeding the warning line and the forecast made contingency measures.
④Dillon also established a model to study the development trend of nutrients, the medium and long-term TN, TP average concentration in the forecast, Dillon found that although the model no short-term forecast accuracy of high precision, but to meet the trend of the accuracy of forecasts, according to water use Model environmental capacity in the medium and long-term forecast on the basis of Wanning Reservoir TN, TP reduction plan, the final scientific pollution prevention and control measures.
随着万宁水库整个流域开发活动的加剧,必然带来大量的对其生态环境不利的问题。因此掌握水体富营养化规律,并对其进行准确预测则显得尤为重要。本文在总结和借鉴前人研究成果的基础上,主要就以下几方面进行了研究:
①本文对万宁水库的水质现状进行了调查研究、收集资料等工作。用因子分析法筛选和分析收集到的资料,找出了影响水库富营养化的主要因子:水温、总磷、总氮、溶解氧、叶绿素a;绘制出万宁水库主要水质参数的分布规律曲线。
②从水库富营养化水平短期预测和营养盐长期预测的角度建立了万宁水库富营养化预测系统。
③为了准确、有效的预测预警,进行水库富营养化短期预测,分析现有预测模型的长处与不足,结合前面分析的数据规律选取了最优模型-BP神经网络,用自适应遗传算法对BP神经网络的不足予以修正。
通过因子分析分析找出预测指标-叶绿素a、水温、总磷、总氮、溶解氧作为模型输入,下月叶绿素a作为输出建立自适应GABP神经网络数学模型。
用Matlab7.0编写程序,采用取水点2000-2005年月平均监测数据对模型进行训练,发现模型拟合度、泛化能力较好:神经网络经865次学习后,误差达到预设精度0.0001,运行总时间仅为36.5770s;拟合值与实测值之间的相关关系系数(R)等于0.999。
模型与没经过自适应遗传算法优化的BP神经网络的训练结果比较发现:自适应学习速率BP神经网络经27950次学习后,误差才收敛到0.001857,除了耗费大量的时间,精度仍不满足要求,两者误差相差一个数量级。
采用修正后的2006年月均监测数据对训练后的自适应GABP模型进行实际预测,最不利的相对误差仅为-11.4%,只有一个相对误差绝对值大于10%,平均相对误差0.172%,预测精度较高,可作为水库富营养化水平预测的依据。
利用自适应GABP的短期预测的精度,对水库富营养化水平预测预警,经过分析,警报点选取为叶绿素a的浓度0.004mg/L,并对超过警报点的预测提出应急措施。
④同时还建立了狄龙模型来研究水库营养盐中长期变化趋势,对中长期TN、TP年平均浓度进行预测,发现狄龙模型精度虽然没有富营养化短期预测精度高,但能够满足中长期预测的精度要求;然后利用经典水质模型计算环境容量,在上述中长期预测的基础上得出万宁水库TN、TP的削减计划,最后提出污染防治措施。
Nowadays the water quality is of gradual shortage, the protection of drinking water is imminent. Eutrophication has become a major global water environment, the global economy maintained rapid growth, has brought increasingly serious problem of environmental pollution. The world’s fast economical development has resulted in more and more seriously Environmental pollution problems such as water eutrophication of lakes and reservoir. In the areas of lots of human activities with the eutrophication is speeding up, the water quality is getting so worse and worse that the development of society and economy is limited. Located in Wanning City in Hainan Province, Wanning Reservoir was built in 1959. It provides services of irrigation, the main source of drinking water, power generation, flood control, fish and so on.
As the Wanning reservoir basin-wide development activities intensify, a large number of ecological environment of negative issues will inevitably be brought about. Therefore it is particularly important to master of eutrophication and predict it accurately. In this paper, to sum up and learn research results from their predecessors, the following points were mainly studied:
①The progress and research methods of the eutrophication is systematically expounded.
②from short-term and long-term forecasts predict the perspective of the establishment of the Wanning reservoir eutrophication forecast management system.
③to accurate and effective early-warning forecast for short-term forecasting, prediction analysis of existing strengths and weaknesses of the model, combining the analysis of the data in front of selecting the optimal model - BP neural networks, genetic algorithms using the BP neural network to be less than that .
In this paper, the Wanning reservoir water quality status of the investigations and studies, collection of information are done. Use factor analysis screening and analysis of the information gathered to identify the impact of eutrophication the main factor Reservoir: water temperature, total phosphorus, total nitrogen, dissolved oxygen, chlorophyll a; drawn Wanning Reservoir water quality parameters of the distribution curve, analyzed Eutrophication-made reservoir, analysis of the causes of eutrophication: in Wanning reservoir in recent years nitrogen, phosphorus have been exceeded the standards; serious Wanning reservoir in the tropical oceans of the monsoon climate, the perennial water temperature higher, the light intensity; slow water flow.
Analysis of the strengths of the existing prediction model and the lack of analysis of the data in front of selecting the optimal model - BP neural networks, genetic algorithms using the BP neural network to be less than that.
Through the mechanism of eutrophication forecast to identify indicators - chlorophyll a, water temperature, total phosphorus, total nitrogen, dissolved oxygen, as a model input, next month as output Chl-a neural network to establish adaptive GABP mathematical model.
With Matlab7.0 preparation procedures, use of water years 2000-2005, an average of monitoring data on the model of training for training and found fitting model, the ability to better generalization: neural networks, the 865 study, the error to a default Accuracy of 0.0001, run total time was 36.5770s; fitting and measured the relationship between the coefficient (R) is 0.999.
Adaptive GABP model and BP neural network optimized not by adaptive genetic algorithm training results found: adaptive learning rate BP neural network by 27,950 after learning of the error before converging to 0.001857, in addition to a significant amount of time, not accuracy meet the requirements of error of the difference from an order of magnitude.
After adaptive GABP model training, forecast by the 2006 years of amended monitoring data: the biggest relative error was -11.4%, only a relative error of absolute value greater than 10 percent, the average relative error of 0.172 percent, Forecast for high precision, the reservoir can be used as the basis for the forecast state of eutrophication.
Adaptive GABP use of short-term forecast accuracy, forecast early warning of water quality, after analysis, warning line selected as the concentration of chlorophyll a 0.004 mg / L, exceeding the warning line and the forecast made contingency measures.
④Dillon also established a model to study the development trend of nutrients, the medium and long-term TN, TP average concentration in the forecast, Dillon found that although the model no short-term forecast accuracy of high precision, but to meet the trend of the accuracy of forecasts, according to water use Model environmental capacity in the medium and long-term forecast on the basis of Wanning Reservoir TN, TP reduction plan, the final scientific pollution prevention and control measures.
引文
[1]黄文钰,舒金华.中国主要水库富营养化评价[J].贵州环保科技, 1997: 12～16.
[2]王占生,刘文君著.微污染水源饮用水处理[M].北京:中国建筑工业出版社.
[3]金士博著.水环境数学模型[M].北京:中国建筑工业出版社, 100-101.
[4]郭劲松,王红,龙腾锐.水资源水质评价方法分析与进展[J].重庆环境科学, 1999, 21(6): 40-42.
[5]彭进新,陈慧君.水质富营养化与防治[M].北京:中国环境科学出版社.
[6]金相灿.湖泊富营养化控制管理技术[M].北京:化学工业出版社.
[7]李炜.环境水力学进展[M].武汉:武汉电力大学出版社.
[8] Sterner, R.W.,J.P. Grover. Algal Growth in Warm Temperate Reserviors:Kinetie Examination of Nitrogen,temperate,Light and other Nutrients[J]. Water Research, 1998, 32: 3539.
[9] C, D.R. Nuyrient Limitation in the sea:Dynamics,Identification and Signicance[J]. Limnol Oceanoqr, 1967(12): 685-695.
[10] K, R.P.K.,L. L. Primary Production Estimation from Continuous Oxygen Measurements in Relation to External Nutrient Input [J]. Water Research, 1996, 30(3): 625-643.
[11]全为民,严力蛟,虞左明.湖泊富营养化模型研究进展[J].生物多样性, 2001, 9(2): 168-175.
[12] A, J.M. An exploratory integrated model to assess management of lake eutrophication[J]. Ecological Modelling, 2001, 140: 111-124.
[13]刘玉生,唐宗武,韩梅.滇池富营养化生态动力学模型及其应用[J].环境科学研究, 1991, 4(6): 1-8.
[14]蔡启铭.太湖环境生态研究[M].北京:气象出版社.
[15]饶群,芮孝芳.完全混合系统总磷随机模型研究[J].水科学进展, 2002, 13(1): 21-25.
[16] Hassan, H.,K. Hanaki,T. Matssuo. A modeling approach to simulate impact of climate change in lake water quality phytoplankton growth rate assessment [J]. Wat. Sci. Tech, 1998, 37(2): 177-185.
[17] J.M.Malmaeus, et al. Lake phosphorus dynamics and climate warming:A mechanistic model approach[J]. Ecological Modelling, 2006(190): 1-14.
[18]金相灿.湖泊富营养化调查规范[M].北京:中国环境出版社, 11-25.
[19]郑涛,穆环珍,黄衍初.非点源污染控制研究进展[J].环境保护, 2005, 2: 31-34.
[20] Reisenhofer, E.,A. Picciotto,a.D. Li. A factor analysis approach to the study of the eutrophication of a shallow, temperate lake[J]. Elsevier Science B.V. , 2000, 306(1): 99-106.
[21]徐振邦,娄元仁.数学地质基础[M].北京:北京大学出版社, 106-108.
[22]卢文喜.长春南湖水质成分的因子分析[J].世界地质, 2003, 18(3): 61-65.
[23]赵新华.水库水质参数预测和富营养化预警[M].天津:天津大学出版社, 13-14.
[24] R.Maier, H.,G. C.dandy. Neural Network Based Modeling of Environmental Variables :A Systematic Approach[J]. Mathematical and Computer Modeling, 2001, 33: 669-682.
[25]罗定贵.基于MATLAB实现的ANN方法在地下水质评价中的应用[J].北京大学学报, 2004, 40(2): 296-302.
[26]卢纹岱. SPSS for Windows统计分析(第2版)[M].北京:电子工业出版社, 413-421.
[27]屠清瑛.巢湖富营养化研究[M].合肥:中国学术科技大学.
[28] S, N.K.,P. K. Gradient Methods for the optimization of dynamical system containing Neural Networks[J]. Neural Netwoks, 2005, 3(2): 252-262.
[29]郭宗楼.人工神经网络在环境质量评价中的应用[J].武汉水利电力大学学报1997, 4(1): 40.
[30]李祚泳. BP网络用于水质综合评价方法的研究[J].环境工程, 2006, 13(2): 22-24.
[31] B, M.,R. J. An introductoon using Neural Network approach to Multiobjective Optimization[J]. Neural Netwoks, 2003, 4(1): 11-14.
[32]袁曾任.人工神经元网络及其应用[M], 66-68.
[33]焦李成.神经网络的应用与实现[M].西安:西安电子科技大学出版社.
[34]吴正茂.神经网络与遗传算法在煤矿智能监控系统中的预测模型研究[M].重庆:西南大学出版社, 11-12.
[35] hmad, S.s.A. Performance of stochastic approaches for forecastting river water quality[J]. Water Resource, 2001, 35(18): 4261-1022.
[36] al, F.R.e. Artificial neural network approach for modeling and prediction of algal blooms[J]. Ecological Modelling, 2005, 146: 139-142.
[37] R.Maier, H.,G. C.Dandy. Understanding the behavior and optimizing the performance of backpropagation neuralnetworks an empirical study[J]. Environmental Modeling &Software, 2003, 13: 179-191.
[38] R.Maier, H. The use of artificial neural networks for the prediction of water quality parameters[J]. Water Resources Research, 2006, 32(4): 1013-1022.
[39]李莹.自适应神经网络在水质预测建模中的应用[J].系统工程, 2001, 19(1): 89-93.
[40]闵惜林,刘国华.人工神经网络结合遗传算法在建模和优化中的应用[J].计算机应用研究, 2002, 1: 79-80.
[41]阎平凡,张长水.人工神经网络与模拟进化计算[M].北京:清华大学出版社.
[42]韩正宇.设备检测中应用基因算法和神经网络[J].西安交通大学, 1998.
[43] E., G. Genetic algorithms in search,optimization and machine learning[J]. Addision WesleyPublishing Company,Inc,2003, 2005.
[44]柏熙.神经网络与遗传算法在煤矿智能监控系统中的预测模型研究[J].西南大学硕士学位论文, 2006.
[45]肖专文,徐日庆.求解复杂工程优化问题的一种实用方法[J].水利学报, 2005, 2: 8-11.
[46]葛红.用遗传算法训练基于神经网络结构的控制器[J].华南师范大学学报(自然科学版), 2006, 1: 24-26.
[47]蒋宗礼.人工神经网络导论[M].北京:高等教育出版社.
[48] Nielsen, R.H. Neurocomputing,Reading[M].
[49]周明,孙树栋.遗传算法原理及应用[M].北京:国防工业出版社.
[50]陈国良.遗传算法及应用[M].北京:人民邮电出版社.
[51] S., J. Parallel Genetic Algorithms:Theory and Applications[M]: ISO Press.
[52]郭晓婷,朱岩.基于遗传算法的进化神经网络[J].清华大学学报, 2000, 40(10): 116-118.
[53]吕锋,石敏.基于GA的BP神经网络模型的研究及应用[J].武汉水利大学学报, 2002(2): 63-66.
[54]刘恒. BP神经网络在千岛湖水体富营养化变化预测中的应用[J].浙江大学硕士毕业论文, 2007: 42.
[55]李志刚.自适应遗传算法的人工神经网络在大坝安全监测中的应用[J].武汉大学硕士论文, 2005: 41-42.
[56]刘涛.水库水质参数预测和富营养化预警[J].天津大学学位论文, 2005: 55-56.
[57]张涛.滇池草海蓝藻清除应急措施总体方案[J].云南环境科学, 2000, 19(1): 38-39.
[58]刘鸿亮,金相灿.湖泊底泥环境疏浚工程技术[J].中国工程科学, 1999, 1(1): 81-84.
[59]黄真理,李锦绣.中国环境水力学[J].中国水利水电出版社, 2002.
[60]赵显波.蘑菇湖水库水环境容量总量控制研究[J].环境工程学报, 2007, 5(11): 12-13.
[2]王占生,刘文君著.微污染水源饮用水处理[M].北京:中国建筑工业出版社.
[3]金士博著.水环境数学模型[M].北京:中国建筑工业出版社, 100-101.
[4]郭劲松,王红,龙腾锐.水资源水质评价方法分析与进展[J].重庆环境科学, 1999, 21(6): 40-42.
[5]彭进新,陈慧君.水质富营养化与防治[M].北京:中国环境科学出版社.
[6]金相灿.湖泊富营养化控制管理技术[M].北京:化学工业出版社.
[7]李炜.环境水力学进展[M].武汉:武汉电力大学出版社.
[8] Sterner, R.W.,J.P. Grover. Algal Growth in Warm Temperate Reserviors:Kinetie Examination of Nitrogen,temperate,Light and other Nutrients[J]. Water Research, 1998, 32: 3539.
[9] C, D.R. Nuyrient Limitation in the sea:Dynamics,Identification and Signicance[J]. Limnol Oceanoqr, 1967(12): 685-695.
[10] K, R.P.K.,L. L. Primary Production Estimation from Continuous Oxygen Measurements in Relation to External Nutrient Input [J]. Water Research, 1996, 30(3): 625-643.
[11]全为民,严力蛟,虞左明.湖泊富营养化模型研究进展[J].生物多样性, 2001, 9(2): 168-175.
[12] A, J.M. An exploratory integrated model to assess management of lake eutrophication[J]. Ecological Modelling, 2001, 140: 111-124.
[13]刘玉生,唐宗武,韩梅.滇池富营养化生态动力学模型及其应用[J].环境科学研究, 1991, 4(6): 1-8.
[14]蔡启铭.太湖环境生态研究[M].北京:气象出版社.
[15]饶群,芮孝芳.完全混合系统总磷随机模型研究[J].水科学进展, 2002, 13(1): 21-25.
[16] Hassan, H.,K. Hanaki,T. Matssuo. A modeling approach to simulate impact of climate change in lake water quality phytoplankton growth rate assessment [J]. Wat. Sci. Tech, 1998, 37(2): 177-185.
[17] J.M.Malmaeus, et al. Lake phosphorus dynamics and climate warming:A mechanistic model approach[J]. Ecological Modelling, 2006(190): 1-14.
[18]金相灿.湖泊富营养化调查规范[M].北京:中国环境出版社, 11-25.
[19]郑涛,穆环珍,黄衍初.非点源污染控制研究进展[J].环境保护, 2005, 2: 31-34.
[20] Reisenhofer, E.,A. Picciotto,a.D. Li. A factor analysis approach to the study of the eutrophication of a shallow, temperate lake[J]. Elsevier Science B.V. , 2000, 306(1): 99-106.
[21]徐振邦,娄元仁.数学地质基础[M].北京:北京大学出版社, 106-108.
[22]卢文喜.长春南湖水质成分的因子分析[J].世界地质, 2003, 18(3): 61-65.
[23]赵新华.水库水质参数预测和富营养化预警[M].天津:天津大学出版社, 13-14.
[24] R.Maier, H.,G. C.dandy. Neural Network Based Modeling of Environmental Variables :A Systematic Approach[J]. Mathematical and Computer Modeling, 2001, 33: 669-682.
[25]罗定贵.基于MATLAB实现的ANN方法在地下水质评价中的应用[J].北京大学学报, 2004, 40(2): 296-302.
[26]卢纹岱. SPSS for Windows统计分析(第2版)[M].北京:电子工业出版社, 413-421.
[27]屠清瑛.巢湖富营养化研究[M].合肥:中国学术科技大学.
[28] S, N.K.,P. K. Gradient Methods for the optimization of dynamical system containing Neural Networks[J]. Neural Netwoks, 2005, 3(2): 252-262.
[29]郭宗楼.人工神经网络在环境质量评价中的应用[J].武汉水利电力大学学报1997, 4(1): 40.
[30]李祚泳. BP网络用于水质综合评价方法的研究[J].环境工程, 2006, 13(2): 22-24.
[31] B, M.,R. J. An introductoon using Neural Network approach to Multiobjective Optimization[J]. Neural Netwoks, 2003, 4(1): 11-14.
[32]袁曾任.人工神经元网络及其应用[M], 66-68.
[33]焦李成.神经网络的应用与实现[M].西安:西安电子科技大学出版社.
[34]吴正茂.神经网络与遗传算法在煤矿智能监控系统中的预测模型研究[M].重庆:西南大学出版社, 11-12.
[35] hmad, S.s.A. Performance of stochastic approaches for forecastting river water quality[J]. Water Resource, 2001, 35(18): 4261-1022.
[36] al, F.R.e. Artificial neural network approach for modeling and prediction of algal blooms[J]. Ecological Modelling, 2005, 146: 139-142.
[37] R.Maier, H.,G. C.Dandy. Understanding the behavior and optimizing the performance of backpropagation neuralnetworks an empirical study[J]. Environmental Modeling &Software, 2003, 13: 179-191.
[38] R.Maier, H. The use of artificial neural networks for the prediction of water quality parameters[J]. Water Resources Research, 2006, 32(4): 1013-1022.
[39]李莹.自适应神经网络在水质预测建模中的应用[J].系统工程, 2001, 19(1): 89-93.
[40]闵惜林,刘国华.人工神经网络结合遗传算法在建模和优化中的应用[J].计算机应用研究, 2002, 1: 79-80.
[41]阎平凡,张长水.人工神经网络与模拟进化计算[M].北京:清华大学出版社.
[42]韩正宇.设备检测中应用基因算法和神经网络[J].西安交通大学, 1998.
[43] E., G. Genetic algorithms in search,optimization and machine learning[J]. Addision WesleyPublishing Company,Inc,2003, 2005.
[44]柏熙.神经网络与遗传算法在煤矿智能监控系统中的预测模型研究[J].西南大学硕士学位论文, 2006.
[45]肖专文,徐日庆.求解复杂工程优化问题的一种实用方法[J].水利学报, 2005, 2: 8-11.
[46]葛红.用遗传算法训练基于神经网络结构的控制器[J].华南师范大学学报(自然科学版), 2006, 1: 24-26.
[47]蒋宗礼.人工神经网络导论[M].北京:高等教育出版社.
[48] Nielsen, R.H. Neurocomputing,Reading[M].
[49]周明,孙树栋.遗传算法原理及应用[M].北京:国防工业出版社.
[50]陈国良.遗传算法及应用[M].北京:人民邮电出版社.
[51] S., J. Parallel Genetic Algorithms:Theory and Applications[M]: ISO Press.
[52]郭晓婷,朱岩.基于遗传算法的进化神经网络[J].清华大学学报, 2000, 40(10): 116-118.
[53]吕锋,石敏.基于GA的BP神经网络模型的研究及应用[J].武汉水利大学学报, 2002(2): 63-66.
[54]刘恒. BP神经网络在千岛湖水体富营养化变化预测中的应用[J].浙江大学硕士毕业论文, 2007: 42.
[55]李志刚.自适应遗传算法的人工神经网络在大坝安全监测中的应用[J].武汉大学硕士论文, 2005: 41-42.
[56]刘涛.水库水质参数预测和富营养化预警[J].天津大学学位论文, 2005: 55-56.
[57]张涛.滇池草海蓝藻清除应急措施总体方案[J].云南环境科学, 2000, 19(1): 38-39.
[58]刘鸿亮,金相灿.湖泊底泥环境疏浚工程技术[J].中国工程科学, 1999, 1(1): 81-84.
[59]黄真理,李锦绣.中国环境水力学[J].中国水利水电出版社, 2002.
[60]赵显波.蘑菇湖水库水环境容量总量控制研究[J].环境工程学报, 2007, 5(11): 12-13.