【作者】 李荣昉；

【导师】 丁永生；

【作者基本信息】 东华大学 ， 模式识别与智能系统， 2013， 博士

【摘要】 我国是一个水资源相对贫乏的国家,自古以来旱灾就是主要自然灾害之一。为解决有限的供水能力和日益增长的需水要求之间的矛盾,必须研究多目标约束条件下水资源高效利用的科学方法,寻找合理调配水源与用水的时空配置方法,达到利用有限的水资源促进经济、社会的持续发展和保护环境健康,实现水资源的可持续利用的目标。水资源的科学合理配置涉及到来水和用水两个过程,除需要解决两个过程的预测问题外,更重要的是需要解决两个过程的时空及要素匹配,以便于应用多目标优化理论与技术,实现来水过程与用水过程的协同优化。为此,本论文参考协同学的相关理论和方法,从构造基于流域水量供需协同优化调度理念的智能系统体系结构入手,对来水过程和用水过程的多因素模拟方法与计算两个过程的协同优化算法等研究成果进行了系统化的论述,并以江西省抚河流域为研究原型,通过与不同方法的对比分析检验了主要研究成果。本论文的主要工作成果如下：(1)制定多级水量协同优化调度参与对象和原则,提出流域多级水量协同优化调度的框架和流程,分析了水量调度系统的功能需求和设计目标的基础,并对研究区域的来水过程和需水过程进行辨析,提出水量调度的依据。(2)系统构建流域控制断面最小需水流量计算模型。依托现有水文站、雨量站、水库和大型灌区取水口,遵循干流与支流相结合的原则,确定流域控制断面和用水区。根据水量分配协商确认方案计算各用水区河道外用水量,利用流域水文站点实测数据和水功能区划资料计算各用水区生态环境需水量,在此基础上,结合区间来水,构建流域各控制断面最小生态需水量计算方法。(3)提出了基于动态免疫克隆粒子群算法的产汇流动态模型。通过实验函数的测试了该算法的优势和用处,分析了算法的快速收敛过程。新安江产汇流计算模型,提供了供水数据处理的途径,但模型的多因子调试和模型的准确性,很难处理。应用提出的基于免疫克隆粒子群算法的动态产汇流参数调配的手段和计算方法,通过改变不同时期的新安江模型参数,使新安江日模型的模拟精度有较大的提高,表明该计算方法具有可行性。(4)基于数据驱动建模和GRNN理论,提出了河流径流的中长期、短期预测模型,并通过对抚河流域洪门水库区域三个主要水文站月径流序列和10天径流序列的预测结果与BPNN模型对比,检验了动态GRNN径流预测模型的拟合精度和预测的可靠性。(5)确定了抚河流域协同水量优化调度的原则,并构建水量调度模型,包括目标函数、约束条件、模型的求解和调度启动条件；基于供需模型的构建,分析不同来水条件下的抚河流域供需状况,构建了协同水量优化调度方法,在非汛期水量协同优化调度方法的基础上,模拟了不同频率来水条件下水量调度结果,验证该方法的可行性。(6)非汛期水量调度辅助决策平台构建。以流域控制断面最小需水流量计算方法、动态供水量预测模型和多级流域水量协同优化调度模型为基础,使用C#语言在Visual Studio2005平台上开发了非汛期水量调度辅助决策系统平台,介绍了水量调度系统平台的运行设备及其运行环境,数据信息流过程。同时提出水量调度专家知识库技术,介绍了水量调度专家知识库查询原理和流程,并验证该技术的可靠性,进而为水量调度方案编制提供支撑。更多还原

【Abstract】 Water resources are relatively poor in China. Drought has been one of major natural disasters in China since ancient times. In order to solve the contradiction between limited water supply capability and ever-increasing water demand, scientific methods have to be found to utilize water resources with high efficiency under multi-objective constraint conditions. Scientific and reasonable methods used to allocate water source and water use, both in spatial and temporal scale, have to be explored. The objectives are to reach the goals of promoting healthy and sustainable development of society, economy and environment with the very limited water resources, and utilizing the water resources in a sustainable way.Scientific and reasonable allocation of water resources are related to two processes:the water supply and the water usage. In addition to solve the problem of water predicting in these two processes, the spatial-temporal and elements matching should be solved in priority. The collaborative theory is a optimal technology and it can be obtained the optimal scheme in a multi-objective optimization and reach the best matching in the application of water supply and the water usage. Using relevant theories and methods of synergetics as references, the structure of intelligent system was constructed on the basis of idea of collaborative optimization water dispatching in river basin. Multiple-factor simulation method and calculation for water supply and water use, and collaborative optimization algorithm have been discussed. A case study was conducted in Fuhe River Basin in Jiangxi Province. Major results and conclusions are verified by means of different methods of analysis. Major contents in this thesis are:(1) The collaborative optimization scheduling and principle for the multi-level water is established, the framework and process of a collaborative optimization of multi-level water dispatching are proposed by analyzing function requirements and design objectives of water dispatching system, In accordance with the results of analyzing the two processes of water supply and water usage in the studied area, the basis for water dispatching scheme is proposed.(2) The water requirement model of the minimum controlling for basin control section is constructed. The basin controlling section and water usage area are decided according to information from relevant hydrologic stations, precipitation stations, water intake of reservoirs and large-scale irrigation areas along the main streams and tributaries in the basin. According to the water allocation scheme, the district river water is calculated, then the water requirement of ecological environment is calculated on basis of the data of the hydrological stations measure and water function materials. After that, combining the interval inflow with river ecological water requirements, a calculation method of each control section is constructed. (3) A dynamic model of concentration and runoff based on immune clonal particle swarm optimization algorithm is proposed. The astringency of the proposed algorithm is analyzed by some functions experiments in detail and the accuration and the reliability is tested. Xin’An River model provides a calculation way for water supply, however, it is difficult to reach accuracy in practice. Appling the proposed dynamic immune clonal particle swarm algorithm to select these parameters in the Xin’An River model adaptively in different periods, the precision of simulation model of Xin’An River has greatly improved, the result shows that the calculation method is feasible.(4) Based on the theory of the data driven and the generalized regression neural network (GRNN), a dynamic long-term and short-term river runoff predictor is proposed The fitting accuracy and reliability of the monthly and10-days predicting model are tested and verified in three major monthly runoff series of hydrological station in Hongmen Reservoir Area in the Fuhe River Basin by comparing with Back Propagation Neural Network (BPNN). The experiment result shows the dynamic GRNN model have a high fitting precision and reliability.(5) The principles used in collaborative optimized water dispatching in Fuhe river basin are formulated. A model used in allocation of water amount is established. The objective function, constraint conditions, solution of model and dispatching startup conditions are determined. On the basis of water supply-demand model, different scenarios of water supply and demand in Fuhe River basin are analyzed under different conditions of water inflow. Methods used in collaborative optimized water dispatching are established. Based on the methods of collaborative optimized water dispatching in non-flooded seasons, scenarios of water dispatching under different frequencies of water supply are simulated with an aim of verifying the feasibility of the model.(6) The platform of decision supporting system used for water dispatching in non-flooded seasons is developed. This platform is constructed using calculating methods for the minimum controlling water requirement for basin control section, dynamic model to predict water supply, and collaborative optimized water dispatching model for multi-level river basin. The platform of decision supporting system used for water dispatching in non-flooded seasons was developed with C#on Visual Studio2005. The running equipment of water dispatching system and their running environment are introduced. The data flow process is also described. Technology of expert’s knowledge base used in water dispatching is studied. The process and inquiry principle of expert’s knowledge base used in water dispatching are illustrated. Also, the reliability of this technology is verified. And then, this technology is used in the formulation of water dispatching scheme.更多还原