节点文献

跨界河流水污染应急决策支持系统研究

Study on Emergency Decision Supporting System for Trans-Boundary River Pollutions

【作者】 郭亮

【导师】 王鹏;

【作者基本信息】 哈尔滨工业大学 , 环境科学与工程, 2011, 博士

【摘要】 与非跨界河流污染相比,跨界河流突发性水污染事故普遍存在特征污染物不易光解、降解,污染强度大、跨度长,波及范围广、时间久的特点。针对我国跨界河流污染风险管理与控制实践中存在的问题,在对国内外典型流域跨界环境污染案例与环境风险管理机制调研的基础上,以典型跨界流域(跨省、国界)松花江流域为研究对象,开展了跨界河流水污染应急决策支持系统研究。将跨界河流污染划分为常规污染与非常规突发污染,基于数据驱动模型研究河流常规污染预测技术,应用MATLAB软件建立了基于人工神经网络(ANN)技术的常规水质预测模型,通过对模型近期和远期预测数据进行分析得到:近期平均预测误差是3.82%,远期平均预测误差是4.74%,即模型对近期的预测效果要明显好于对远期水质的预测效果。对于突发污染高浓度污染团长时间、大尺度传输问题,基于机理模型—地表水模拟系统(Surface Water Modeling System,SMS)推演污染演替过程。同时基于不确定性原理构建二维随机水质模型,进一步研究数据稀缺条件下突发水污染事故污染带尺寸和目标断面浓度时间线的预测与分析。利用相关实测数据采用确定性和不确定性方法分别对模型进行了参数率定,得到了横向与纵向传质系数的概率分布和最大频率;污染带长度和宽度的模拟验证的平均误差小于15%,表明模型可较准确模拟计算河流污染带。针对环境监管部门对跨界污染事件应急处置技术预案缺乏科学有效评估方法问题,和对国内外跨界水环境污染事件进行分类梳理基础上,设计构建了基于案例与规则联合推理的跨界环境污染应急预案库。利用层次分析法(AHP)建立案例评价指标体系,计算指标权重,提出包括问题描述、解决方案和结果评价框架的案例表示标准模型,比较分析了常用的数学算法,采用改进的欧氏距离法用于案例匹配以消除指标间数量级差异引起的计算问题,并利用决策树方法建立规则推理机制以完善预案生成过程。最后基于盲数理论建立预案可信度评价算法,并对松花江硝基苯污染案例进行验证。基于博弈理论对跨界河流污染过程中,地方政府间竞争与合作这一不完全信息动态博弈过程进行分析,辨识与定义决策过程中的博弈要素,构建博弈双方收益矩阵,遵从纳什均衡概念寻求群体决策策略均衡点,建立跨界流域污染群体决策模式。基于“情景-应对”型应急管理理论与多智能体(Multi-Agent)技术研究了跨界风险协同决策机理,将流域上下游众多部门视为独立的智能体(Agent),采用带权分配的合同网协议(CNP)作为Agent间的协作机制,建立多智能体协同决策模型。将流域跨界污染问题视为复杂问题求解过程,进行分解和重构的过程,基于分布式并行处理的设计思想,由传统的群体一致性算法处理每个子问题的结果,消解多智能体间子预案冲突问题。集成人工智能和软件工程技术,基于WebGIS技术建立了跨界河流水污染应急决策支持系统,并设计了相关数据库,以2005年松花江流域硝基苯污染事件为例,反演历史污染案例,验证河流长程污染模拟、快速预案生成及决策会商模块,实现了环境风险信息的通报与共享、流域水污染事故的应急响应处理与法律法规查询等功能,为有效控制松花江流域跨界污染提供群体决策支持。

【Abstract】 Compared with normal river pollution, the pollutions, in the sudden water pollution accidents of trans-boundary river, have the characteristics: photolysis-resistent and biodegradation-resistent, heavy damage, long duration and large ranges. Due to the problems in the practice of risk management and control of trans-boundary river pollution, 362 environmental pollution incidents of past five years have been collected, especially the four incidents have been emphasized on, the water pollution incident in Songhua River in 2005 and in 2010, the two water pollution incident over the boundary of Jiangsu and Shandong in 2009, the domestic and foreign typical watershed cross-border environment pollution case and environmental risk management mechanism are research basis, typical trans-boundary basin (inter-provincial, borders) the songhua river basin are the research object we studied on trans-boundary river pollution emergency decision support system technology.Trans-boundary river pollution was divided into conventional pollution and unconventional sudden pollution. Conventional pollution prediction technology was studied based on data-driven module, applying MATLAB to establish conventional water quality prediction model, Through recent and long-term prediction model analysis of the data available, it conclude that recent average prediction error is 3.82%, long-term average prediction error is 4.74%, which means forecast effect of recent significantly better than that of long-term. In order to predict high concentration pollution plume during emergency with long-time transport and large-scale transmission, the process was modeled by SMS(Surface Water Modeling System) based on mechanism-driven module, meanwhile, constructing two-dimensional random water quality model based on uncertainty principle to research sudden water pollution accidents with data scarce condition, and slick size and concentration of the time line target section was computational predicted and analyzed. By using the related experimental data validate the model parameters set by the means of certainty and uncertainty method separately, obtained that the rate of transverse and longitudinal mass transfer coefficient probability distribution and maximum frequency, Slick length and width of the average error of the analog verification less than 15%, indicating that the model can accurately simulated calculation the polluted plume.To solve the problem of environmental authorities, for the lack of scientific and effective technical assessment on trans-boundary pollution incident emergency plans, Casebase based on CBR/RBR was designed on the basis of analyzing trans-boundary water environmental pollution events happened in domestic and abroad. Using the analytic hierarchy process (AHP) to build case evaluation index system, calculation index weight and proposed including problem description, solutions and results evaluation framework. Comparing and analyzing common mathematical algorithm and the standard model, adopting an improved Euclidean distance method to eliminate index for case matching between the magnitude of differences cause, using decision tree method to establish rule reasoning mechanism to perfect plan formation process. Based on blind number theory, established the credibility assessment algorithm of the plan and the case of nitrobenzene pollution of Songhua river was used for verification.Non-complete information dynamic game process between the local governments was analyzed based on game theory during the pollution happened in in the trans-boundary basin. Identification and definition of the game elements in decision process was made and return matrix of the game was also constructed. Following Nash equilibrium concept for group decision strategy equilibrium, establishing trans-boundary basin pollution group decision model.Based on the scenario-responses emergency management theory and of multi-Agent technology, the collaborative decision-making mechanism in trans- boundary emergency was researched and multitudinous departments in upstream and downstream of river basin was designed as an Agent, modified contract nets protocol (CNP) was chosen as Agent collaboration mechanism, at last, collaborative decision- making of multi-Agent system was established based on the thought of distributed parallel processing design. The conflicts of Agents was solved through the traditional group consistency algorithm processing.Finally, integrating artificial intelligence and software engineering technology, water pollution emergency decision support system and its database and case base was constructed based on WebGIS technology. Case of nitrobenzene pollution happened in Songhua river basin, 2005 was chosen as inversion case to validate river’s long-range pollution simulating, rapid preplan generation and decision-making consultation. The function of environmental risk information bulletin and sharing, water pollution emergency response, laws and regulations inquires was realized for the purpose of effective control of the pollution of Songhua river basin and it provides group decision support for trans-boundary pollution.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络