水厂源水水质监测、预测、数据传输系统研究
|
摘要
国内大部分地表水水厂，近年由于水源污染，水质恶化，富营养化严重，水厂净水系统在常规水处理工艺基础上，正在研究和开发常规强化、预处理和深度处理技术。而水源水质的变化成为决定水处理工艺及操作运行条件的重要因素。因此，在保证供水水质的前提下，建立源水水质监测、预测及数据传输系统对于实现水质变化条件下启／停非常规工艺、实时优化及调整水处理工序及工艺参数、实现水厂经济运行有重要的意义。
引滦入津的滦河水近年季节性的藻类高发，给水厂运行带来很大的危害。天津自来水集团公司所属的三个水厂已采用了不同预处理措施以降低藻类对常规水处理工艺的影响。为保证供水水质，实时优化水处理工艺，在本次研究中建立源水水质在线监测与预测系统是研究的重要内容之一。
为建立预测藻类高发的水质模型，收集了芥园水厂水源地—预沉池上游宜兴埠泵站1997—2002年1683组的日监测水质资料(不连续)，在常年监测的指标中叶绿素的检测频率是最高，内插的数据最少，将其确定为藻类高发指示指标。藻类的动态生长变化非常复杂，线性模型过于简单，而具有非线性特征的神经网络是预测叶绿素的适宜方法。通过灰色关联度分析、相关分析、指标聚类分析、通径分析及藻类生长机理分析选择了模型不同的输入变量，据此确定了5种不同预测模型；通过对灰色关联度—神经网络预测模型多次数学试验，选择了适宜模型结构、输入延迟等神经网路模型参数，再将此参数应用于其它预测模型，并对模型进行优选，最终确定指标聚类—神经网络模型是预测滞后3d叶绿素应用模型。其拓扑结构为：6-3-3-1前馈神经网络，输入延迟为6，即要求变量是此前连续6d水质指标，输入变量：叶绿素、浊度、氨氮、水温、pH、总碱度。通过实际验证，叶绿素预测准确率在80％以上。
在天津市两个水厂的水源地(芥园和凌庄)—西河预沉池建立了源水水质在线监测系统。在调研仪表检测原理和生产现状的基础上，完成了仪表选型；在学习数据采集相关软
西安建筑科技大学硕士学位论文
硬件相关知识后，系统经精心设计、安装及在LabView平台上开发出了水质在线监测软件，
所建成系统的运行稳定可靠。
通过GPRS组建监测数据传输网络及利用LabView平台上数据套接技术，经济可靠地
建立了源水远程监测虚拟仪器。现阶段芥园水厂已经实现了源水水质实时监测、数据采集、
存储及处理，减少了源水人工监测频次。论文研究成果为下阶段建立源水水质预警系统奠
定了良好基础。
For surface water polluted seriously, water quality deteriorated sharply, mass organics on raw water and so on recently, many nontraditional treatments would be added to diagrams of water purifications to meet the standard quality of supplied waters on lots of water plants at home. A system of water quality-monitoring and forecasting and transmission on raw water is significant in the operation of nontraditional treatments, real-time optimization and adjustment of diagram and parameters of units of water purification so that the plant run economically and the water supplied reach the relative standard.
The water on Luan River reached to Tianjin has a sharply change which lots of algae can be found on later spring and summer. Plants along the rivers run awful. Three water works under Tianjin water limited company carry out different pre-treatments in order to decrease the impact of algae on units of water purification. For the quality of water supplied and the optimization of water purification's diagram, the system of water quality-monitoring and forecasting and transmission on water of Tianjin's Luan River was built under this study.
1683 non- continuous diary data (1997.8.12-2000.10.3, and 2001.5.1-2002.10.16) about water qualities of Yixingfu pump station where is to the pre- sedimentation tank of Jieyuan water work as one of three plants were collected for water quality model about forecasting the flourishing algae. Since the frequency of chlorophyll measured is the highest on all water parameters and the intervened data of chlorophyll measured is the lowest, chlorophyll would be an indicative parameter of algae. The growth of algae is so complex and the linear model is so simple that the best method for forecasting model about the intensity of chlorophyll is the artificial neural network. The input parameters were selected via analyzing linear correlation, grey correlation, clustering, passaging and the principle of algae growth. According to inputs, forecasting models were divided into 5 groups. Lots of mathematic tests on the forecasting model that the inputs were selected by grey correlation were made so that the preferences of
the neural network model, for instance: mode topology, input delay and so on, were chosen, then the preferences were applied on other model and the optimized artificial neural network was selected among these models, which inputs were chosen by clustering. Its topology, delay, and inputs are 6-3-3-1, 6 and chlorophyll, turbidity, ammonia, water temperature, pH, total alkalinity. The trained model can simulate the volume of chlorophyll the third day by all data 6 days before. The forecasts accurate can reach 80% and meet user's needs.
The system of real-time monitoring on raw water quality was built on the pre-sedimentation tank where the source water of two works comes from at Xihe by author. The automatic instruments were chosen based on the investigation about the measuring principals and current productions. After I have studied the corresponding software and hardware about the system of data collecting, the system was design and install elaborately, then the software on real-time monitoring water quality was programmed on LabView, all these can have the system run reliably.
The remote virtual instrument on source water quality was built economically and reliably via data transmission on GPRS and the Datasocket technique on LabView. At present, Jieyuan water work can monitor real-time its source water and collect, save and treat the monitored water quality data so that the frequency of measurement manually on raw water can be largely decreased. All these ends would be an solid base for the system of source water warning next stage.
引滦入津的滦河水近年季节性的藻类高发，给水厂运行带来很大的危害。天津自来水集团公司所属的三个水厂已采用了不同预处理措施以降低藻类对常规水处理工艺的影响。为保证供水水质，实时优化水处理工艺，在本次研究中建立源水水质在线监测与预测系统是研究的重要内容之一。
为建立预测藻类高发的水质模型，收集了芥园水厂水源地—预沉池上游宜兴埠泵站1997—2002年1683组的日监测水质资料(不连续)，在常年监测的指标中叶绿素的检测频率是最高，内插的数据最少，将其确定为藻类高发指示指标。藻类的动态生长变化非常复杂，线性模型过于简单，而具有非线性特征的神经网络是预测叶绿素的适宜方法。通过灰色关联度分析、相关分析、指标聚类分析、通径分析及藻类生长机理分析选择了模型不同的输入变量，据此确定了5种不同预测模型；通过对灰色关联度—神经网络预测模型多次数学试验，选择了适宜模型结构、输入延迟等神经网路模型参数，再将此参数应用于其它预测模型，并对模型进行优选，最终确定指标聚类—神经网络模型是预测滞后3d叶绿素应用模型。其拓扑结构为：6-3-3-1前馈神经网络，输入延迟为6，即要求变量是此前连续6d水质指标，输入变量：叶绿素、浊度、氨氮、水温、pH、总碱度。通过实际验证，叶绿素预测准确率在80％以上。
在天津市两个水厂的水源地(芥园和凌庄)—西河预沉池建立了源水水质在线监测系统。在调研仪表检测原理和生产现状的基础上，完成了仪表选型；在学习数据采集相关软
西安建筑科技大学硕士学位论文
硬件相关知识后，系统经精心设计、安装及在LabView平台上开发出了水质在线监测软件，
所建成系统的运行稳定可靠。
通过GPRS组建监测数据传输网络及利用LabView平台上数据套接技术，经济可靠地
建立了源水远程监测虚拟仪器。现阶段芥园水厂已经实现了源水水质实时监测、数据采集、
存储及处理，减少了源水人工监测频次。论文研究成果为下阶段建立源水水质预警系统奠
定了良好基础。
For surface water polluted seriously, water quality deteriorated sharply, mass organics on raw water and so on recently, many nontraditional treatments would be added to diagrams of water purifications to meet the standard quality of supplied waters on lots of water plants at home. A system of water quality-monitoring and forecasting and transmission on raw water is significant in the operation of nontraditional treatments, real-time optimization and adjustment of diagram and parameters of units of water purification so that the plant run economically and the water supplied reach the relative standard.
The water on Luan River reached to Tianjin has a sharply change which lots of algae can be found on later spring and summer. Plants along the rivers run awful. Three water works under Tianjin water limited company carry out different pre-treatments in order to decrease the impact of algae on units of water purification. For the quality of water supplied and the optimization of water purification's diagram, the system of water quality-monitoring and forecasting and transmission on water of Tianjin's Luan River was built under this study.
1683 non- continuous diary data (1997.8.12-2000.10.3, and 2001.5.1-2002.10.16) about water qualities of Yixingfu pump station where is to the pre- sedimentation tank of Jieyuan water work as one of three plants were collected for water quality model about forecasting the flourishing algae. Since the frequency of chlorophyll measured is the highest on all water parameters and the intervened data of chlorophyll measured is the lowest, chlorophyll would be an indicative parameter of algae. The growth of algae is so complex and the linear model is so simple that the best method for forecasting model about the intensity of chlorophyll is the artificial neural network. The input parameters were selected via analyzing linear correlation, grey correlation, clustering, passaging and the principle of algae growth. According to inputs, forecasting models were divided into 5 groups. Lots of mathematic tests on the forecasting model that the inputs were selected by grey correlation were made so that the preferences of
the neural network model, for instance: mode topology, input delay and so on, were chosen, then the preferences were applied on other model and the optimized artificial neural network was selected among these models, which inputs were chosen by clustering. Its topology, delay, and inputs are 6-3-3-1, 6 and chlorophyll, turbidity, ammonia, water temperature, pH, total alkalinity. The trained model can simulate the volume of chlorophyll the third day by all data 6 days before. The forecasts accurate can reach 80% and meet user's needs.
The system of real-time monitoring on raw water quality was built on the pre-sedimentation tank where the source water of two works comes from at Xihe by author. The automatic instruments were chosen based on the investigation about the measuring principals and current productions. After I have studied the corresponding software and hardware about the system of data collecting, the system was design and install elaborately, then the software on real-time monitoring water quality was programmed on LabView, all these can have the system run reliably.
The remote virtual instrument on source water quality was built economically and reliably via data transmission on GPRS and the Datasocket technique on LabView. At present, Jieyuan water work can monitor real-time its source water and collect, save and treat the monitored water quality data so that the frequency of measurement manually on raw water can be largely decreased. All these ends would be an solid base for the system of source water warning next stage.
引文
[1] Qing Zhang.Stephen J.Stanley. et. al.. Forecasting raw-water quality parameters for the North Saskatchewan river by neural network modeling[J]. Water. Res.1997,Vo131(9):2340-2350.
[2] Harro bode. Ernst A.Nusch. Advanced river quality monitoring in the RUHR Basin [J] Wat.Sec.Tec. 1999 Vo140. No10. P145-152.
[3] 齐文启，孙宗光，边归国等．环境监测新技术[M]．化学工业出版社．2003
[4] 葛刚．建立水质自动监测系统的思考．干旱环境监测[J]．2000．14(4)：226-229
[5] 吴邦灿．环境监测技术[M]．中国环境科学出版社，1995
[6] 吴邦灿．黄龙．现代环境监测技术[M]．中国环境科学出版社，1999
[7] 吴忠标．吴组成．现代监测[M]．化学工业出版社，2003
[8] Shamshad Ahmad et. al. Performance of stochastic approaches for forecasting river water quality [J].Water. Res., 2001, 35(18):4261-4266
[9] Holger R. Maier et. al. The use of artificial neural networks for the prediction of water quality parameters [J]. Water resources research,1996, 32 (4): 1013-1022
[10] Friedrich Recknagel et. al., Artificial neural network approach for modeling and prediction of algal blooms [J]. Ecological Modeling, 1997, 96: 11-28
[11] 王开章等，灰色模型在大武水源地水质预测中的应用[J]．山东农业大学学报(自然科学版)，2002，33(1)：66-71
[12] 赵雪辉等，灰色系统GM(1，1)残差模型在水质预测中的应用与探讨[J]．干旱环境监测，1997，11(2)：118-120
[13] 李莹等，河流水质的预测模型研究[J]．系统仿学报，2001，13(2)：139—142
[14] Holger R. Maier et. al. Forecasting cyan bacterium Anabaena spp. in the River Murray, South Australia,using B-spline neurofuzzy models[J]. Ecological Modeling, 2001, 146:85-96
[15] Jason Bobbin et. al. Inducing explanatory rules for the prediction of algal blooms by genetic algorithms[J]. Environment International, 2001, 27:237-242
[16] 谢永明．环境水质概论冲国科学技术出版社．1996
[17] 丁训静等．太湖流域水质模拟研究．水资源保护．1998，5(1)：10-14
[18] 马薇纯．陈燕．运用系统工程方法对苏州河进行整治．上海环境科学．1998，17(6)：7-8
[19] 《中国湖泊环境》第一册 金相灿 海洋出版社．2003
[20] 徐景翼 当前供水及水处理面临的问题和对策的探讨2000年中国水协科技委含藻处理研讨会论文
[21] 张华梁 阴沛军 处理高藻滦河水的对策 www.waterchina0com
[22] 秦保平 孙韧等 引滦河道中水质偏碱的起因研究 环境科学研究1999,12(5)：41-42
[23] 飞思科技产品研发中心MATLAB6.5辅助神经网络分析与设计 电子工业出版社2003
[24] Richard J. Roiger. Michael W.Geatz. Data Mining A tutorial-Based Primer[M]. 1st Edition. Pearson Education, Inc.2003.
[25] Holger R. Maier. Grame C. Dandy. Understanding the behavior and optimizing the performance of back-propagation neural networks:an empirical study[J]. Environmental Modeling & Software,1998,13:179-191.
[26] Joseph H.W.Lee, Yan Huang, Mike Dickman, A.W. Jayawardena. Neural network modeling of coastal algal blooms. Ecological Modelling 2003,159:179-201
[27] Fredric M.Ham, Ivica Kostanie. Principles of Neuraleomputing for Science & Engineering[M]. English edition, 2003
[28] Holger R. Maier. Grame C. Dandy. Neural Network Based Modeling of Environmental Variables: A Systematic Approach, Mathematical and Computer Modeling, 2001,33:669-682
[29] Simon Haykin NEURAL NETWORK A Comprehensive foundation (Second Edition)[M] Tsinghua University Press 2001
[30] 刘思峰 郭天榜，党耀国等，灰色系统理论及其应用[M]2000
[31] 邓聚龙 灰色系统基本方法[M] 华中理工大学出版社1987
[32] 邓聚龙 灰色预测与控制[M] 华中理工大学出版社1986
[33] 邓聚龙 灰色控制系统[M] 华中理工大学出版社1985
[34] 汪荣鑫 数理统计[M] 西安交通大学出版社1986
[35] 于秀林，任雪松．多元统计分析[M]中国统计出版社，2001
[36] 袁志发 周静芋等，多元统计分析[M]，科学出版社2002
[37] 范希民等 神经网络和遗传算法在水科学领域的应用[M]中国水利水电出版社2002
[38] 徐景翼 当前供水及水处理面临的问题和对策的探讨 2000年中国水协科技委含藻处理研讨会论文
[39] 张文彤，SPSS11统计分析教程[M]，北京希望电子出版社，2002
[40] 蔡皖东 计算机网络技术[M]．北京科学出版社2003
[41] 彭祖林 梅林 叶超 网络系统集成工程项目投标与施工[M]．北京国防工业出版社2003
[42] 杨乐平 李海涛 赵勇 杨磊 Lab VIEW高级程序设计[M]．北京清华大学出版社2003
[43] 乔瑞萍 林欣 LabVIEW6i实用教程[M]．北京电子工业出版社2002年
[44] 曹玲芝 崔光照 吴刚 现代测试技术及虚拟仪器[M]北京航空航天大学出版社2004
[2] Harro bode. Ernst A.Nusch. Advanced river quality monitoring in the RUHR Basin [J] Wat.Sec.Tec. 1999 Vo140. No10. P145-152.
[3] 齐文启，孙宗光，边归国等．环境监测新技术[M]．化学工业出版社．2003
[4] 葛刚．建立水质自动监测系统的思考．干旱环境监测[J]．2000．14(4)：226-229
[5] 吴邦灿．环境监测技术[M]．中国环境科学出版社，1995
[6] 吴邦灿．黄龙．现代环境监测技术[M]．中国环境科学出版社，1999
[7] 吴忠标．吴组成．现代监测[M]．化学工业出版社，2003
[8] Shamshad Ahmad et. al. Performance of stochastic approaches for forecasting river water quality [J].Water. Res., 2001, 35(18):4261-4266
[9] Holger R. Maier et. al. The use of artificial neural networks for the prediction of water quality parameters [J]. Water resources research,1996, 32 (4): 1013-1022
[10] Friedrich Recknagel et. al., Artificial neural network approach for modeling and prediction of algal blooms [J]. Ecological Modeling, 1997, 96: 11-28
[11] 王开章等，灰色模型在大武水源地水质预测中的应用[J]．山东农业大学学报(自然科学版)，2002，33(1)：66-71
[12] 赵雪辉等，灰色系统GM(1，1)残差模型在水质预测中的应用与探讨[J]．干旱环境监测，1997，11(2)：118-120
[13] 李莹等，河流水质的预测模型研究[J]．系统仿学报，2001，13(2)：139—142
[14] Holger R. Maier et. al. Forecasting cyan bacterium Anabaena spp. in the River Murray, South Australia,using B-spline neurofuzzy models[J]. Ecological Modeling, 2001, 146:85-96
[15] Jason Bobbin et. al. Inducing explanatory rules for the prediction of algal blooms by genetic algorithms[J]. Environment International, 2001, 27:237-242
[16] 谢永明．环境水质概论冲国科学技术出版社．1996
[17] 丁训静等．太湖流域水质模拟研究．水资源保护．1998，5(1)：10-14
[18] 马薇纯．陈燕．运用系统工程方法对苏州河进行整治．上海环境科学．1998，17(6)：7-8
[19] 《中国湖泊环境》第一册 金相灿 海洋出版社．2003
[20] 徐景翼 当前供水及水处理面临的问题和对策的探讨2000年中国水协科技委含藻处理研讨会论文
[21] 张华梁 阴沛军 处理高藻滦河水的对策 www.waterchina0com
[22] 秦保平 孙韧等 引滦河道中水质偏碱的起因研究 环境科学研究1999,12(5)：41-42
[23] 飞思科技产品研发中心MATLAB6.5辅助神经网络分析与设计 电子工业出版社2003
[24] Richard J. Roiger. Michael W.Geatz. Data Mining A tutorial-Based Primer[M]. 1st Edition. Pearson Education, Inc.2003.
[25] Holger R. Maier. Grame C. Dandy. Understanding the behavior and optimizing the performance of back-propagation neural networks:an empirical study[J]. Environmental Modeling & Software,1998,13:179-191.
[26] Joseph H.W.Lee, Yan Huang, Mike Dickman, A.W. Jayawardena. Neural network modeling of coastal algal blooms. Ecological Modelling 2003,159:179-201
[27] Fredric M.Ham, Ivica Kostanie. Principles of Neuraleomputing for Science & Engineering[M]. English edition, 2003
[28] Holger R. Maier. Grame C. Dandy. Neural Network Based Modeling of Environmental Variables: A Systematic Approach, Mathematical and Computer Modeling, 2001,33:669-682
[29] Simon Haykin NEURAL NETWORK A Comprehensive foundation (Second Edition)[M] Tsinghua University Press 2001
[30] 刘思峰 郭天榜，党耀国等，灰色系统理论及其应用[M]2000
[31] 邓聚龙 灰色系统基本方法[M] 华中理工大学出版社1987
[32] 邓聚龙 灰色预测与控制[M] 华中理工大学出版社1986
[33] 邓聚龙 灰色控制系统[M] 华中理工大学出版社1985
[34] 汪荣鑫 数理统计[M] 西安交通大学出版社1986
[35] 于秀林，任雪松．多元统计分析[M]中国统计出版社，2001
[36] 袁志发 周静芋等，多元统计分析[M]，科学出版社2002
[37] 范希民等 神经网络和遗传算法在水科学领域的应用[M]中国水利水电出版社2002
[38] 徐景翼 当前供水及水处理面临的问题和对策的探讨 2000年中国水协科技委含藻处理研讨会论文
[39] 张文彤，SPSS11统计分析教程[M]，北京希望电子出版社，2002
[40] 蔡皖东 计算机网络技术[M]．北京科学出版社2003
[41] 彭祖林 梅林 叶超 网络系统集成工程项目投标与施工[M]．北京国防工业出版社2003
[42] 杨乐平 李海涛 赵勇 杨磊 Lab VIEW高级程序设计[M]．北京清华大学出版社2003
[43] 乔瑞萍 林欣 LabVIEW6i实用教程[M]．北京电子工业出版社2002年
[44] 曹玲芝 崔光照 吴刚 现代测试技术及虚拟仪器[M]北京航空航天大学出版社2004