【作者】 段春青；

【导师】 黄强； 邱林；

【作者基本信息】 西安理工大学 ， 水文学及水资源， 2007， 博士

【摘要】 随着我国人口增加和经济发展,水资源短缺问题日趋严重,灌区地表水与地下水的联合调度与实时调度的地位和作用变得越来越突出。如何能够通过管理手段实现灌区水资源的高效利用成为国内外学者们研究的热点和难点。本论文较为系统地介绍了灌区调度的研究现状与发展趋势,引入现代的优化、预测和决策技术,建立了地表水与地下水联合调度多目标模型,从不确定性和不完备性角度出发,研究了灌区中长期灌溉制度制定问题,并采用了先进的数据挖掘技术建立了灌区实时优化调度模型。论文主要取得了以下成果:(1)将混沌优化算法引入到作物灌溉制度优化设计之中;针对等式约束优化问题的特点,对传统实码遗传算法进行改进,提出了超平面实码遗传算法,并基于此算法建立了灌区灌溉制度优化设计的大系统分解协调模型,经实例验证该模型对求解等式约束的高维非线性问题有较高的效率,能够得出精度更高的典型年优化灌溉制度。(2)将地下水抽蓄变化等效地看作水库的调度,借助地下水运动的模拟技术和混沌优化算法,在考虑地下水动态变化的基础上建立了地表水与地下水联合调度的多目标模型,并采用交互切比雪夫方法对模型进行求解。该方法具有全面性、高效性、实用性和客观性等优点,使得取得决策者偏好这一比较困难的工作变得相当容易,充分发挥了人与计算机的特长。通过建立多水源多目标联合调度模型,将地下水位控制在合理范围内,以期能在最大限度获得效益的基础上实现地下水的可持续利用和避免环境问题的发生。(3)建立了基于作物需水和来水过程的年型判别模型,根据作物需水和天然来水过程的匹配程度定义年份的丰枯程度,避免了根据年总水量定义丰枯常常与作物实际生长情况不符的缺陷。(4)将云理论引入到灌区中长期灌溉制度制定问题的研究中,建立了基于云推理的灌溉制度制定模型,该模型能够同时考虑年型预测中的随机和模糊不确定性,从历史数据中挖掘不确定性知识,在年型的预测和灌溉制度的制定上取得了较好的效果。(5)将信息扩散理论引入到资料较少条件下的灌区中长期灌溉制度制定问题的研究中,建立了基于信息扩散近似推理的灌区灌溉制度制定模型,此模型能在资料较少,信息不完备条件下,尽可能地多利用和挖掘信息,得出比较符合实际的结果。(6)建立了基于遗传程序设计的参考作物腾发量预测模型。借助云预测技术,从纵向阶段降水数据和横向阶段降水比例两个角度,同时进行阶段降水预测分析。与传统预测方法比较,该模型提高了预测的精度。在此基础上,结合渠村灌区资料建立了灌区实时优化调度模型,实现了短期调度与中长期调度的耦合。更多还原

【Abstract】 With rapid economy development and population increase, it is more and more important for irrigation area to take coordinative dispatch of surface water and groundwater and real time optimal regulation. How to realize efficient utilization of water resources in irrigation area by management? This is the hotspot and difficulty in water resources field. This paper establishes the multi-purpose model of coordinative dispatch of surface water and groundwater in irrigation area on basis of the research trend in the world, puts forward methods for forecasting irrigation schedule based on uncertain and incomplete theory, and suggests real time optimal model for irrigation regulation with the technique of data mining. The main results of this paper are as follows:(1) Chaos algorithm is introduced to analyze optimal irrigation schedule for crop. Put forward hyperplane real code genetic algorithm with which optimal irrigation schedule for irrigation area is established. These methods improve greatly the accuracy and speed of calculation for non-linear programming. Better irrigation schedule of typical year can be gotten by these models.(2) Establish multi-purpose coordinative dispatch model of surface water and groundwater with simulation technique of groundwater movement and chaos algorithm. This model takes the dynamic variance of groundwater level into account, and is solved by Chebysher’s method that is efficient, feasible, objective, and comprehensive. By controlling groundwater level properly, sustainable utilization of groundwater can be realized and environmental problem can be avoided.(3) Suggest judgment model based on procedures of precipitation and water demanding of crop. This model defines again the adequate and inadequate extent of year from the matching degree of two procedures, avoiding the shortcoming of traditional judgment method.(4) Based on cloud theory, model for establishing irrigation schedule of middle and long term is brought forth. It can consider random and fuzzy uncertainty in forecast problem, and get better results by mining uncertain knowledge from historical data.(5) Information diffusion theory is used to take research to establishing irrigation schedule too. Model with information diffusion technique can take advantage of more information under the condition of data shortage and get better results than other classic methods.(6) Bring forth evaportranspiration model of reference crop based on genetic programming. Analyze the forecast problem of precipitation in growth phase with cloud model, considering lengthways precipitation value and transverse precipitation proportion. This model with cloud theory can get better results than traditional forecast methods. Based on the forecast model, real time optimal regulation model of irrigation area is put forward, which realizes the coupling between short-range regulation and long-range regulation.更多还原