【作者】 陈晓楠；

【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， 2008， 博士

【摘要】 随着我国经济建设的迅速发展,水资源短缺问题日趋严重,干旱灾害已经成为我国经济可持续发展的瓶颈。在长期与干旱斗争的实践中,人们认识到干旱灾害是不可避免的,只能尽力降低干旱带来的损失。传统干旱灾害的危机管理已经不适用于当前水资源危机的严峻形势,必须实施干旱风险管理,从被动抗旱向积极主动抗旱转变。如何能够有效地对干旱灾害进行风险管理已成为国内外学者们研究的热点和难点。我国是一个农业大国,农业是国民经济的基础,水资源则是农业的命脉,对农业干旱进行系统、深入的风险研究显得十分迫切。在此背景下,本论文在综合国内外相关研究成果的基础上,引入现代的优化、预测和决策技术,建立起农业干旱灾害风险管理体系,从不确定性角度出发,详细地研究了农业干旱灾害的时、空发展规律以及抗旱对策。具体内容包括以下几点:(1)指出干旱定义应考虑水分供需的两个方面,并强调干旱是一种异常现象。基于此思想,完善了干旱及农业干旱的定义。改进后的定义不仅考虑了自然因素,还考虑到人为因素。不但考虑到供水情况,还考虑到需水情况。强调应以正常情况为标准描述干旱,这比用理想情况为标准描述干旱更加符合实际;提出了扩展风险以及扩展自然灾害风险的概念,用灾害预测的影响程度和预测的可靠程度来描述风险。这种风险定义不但扩展了使用范围,而且为决策者提供了更丰富的信息。以此为基础,结合灾害学,建立了农业干旱灾害风险管理体系结构。(2)建立了基于云推理自回归模型的年降水量预测模型、基于信息扩散近似推理的年降水量预测模型。这些模型充分考虑了年降水量之间的随机性和模糊性,能够较好地从历史数据中挖掘知识,描述复杂的非线性关系。与传统预测方法比较,模型能够得到更高预测精度的结果。通过对年降水量规律的研究,实现农业干旱危险性的分析。(3)提出了农业干旱评估应反映出干旱给农业造成的损失大小的思想,并提出模型应方便地对未来干旱程度进行风险评估的主张;将遗传程序设计引入到作物水分生产函数的研究中,通过程序自动寻找最优的函数结构,取得了很好的效果,为作物产量与水分胁迫关系的描述提供了新的思路;建立了两层土壤计算层的农业干旱风险评估的静态、动态模型。利用这些模型不仅能够完成不同时间尺度下的干旱程度的风险评估,而且能较准确地反映农业的损失。(4)提出多种规避、降低农业干旱灾害风险的策略,包括建立起基于群决策的农业干旱预警模型、基于自适应随机惯性权粒子群的农业水资源优化配置模型以及农业水资源承载力模型,对农业干旱灾害实施全面、综合管理,实现水资源的可持续利用。(5)分析农业干旱程度的概率分布规律,提出了使用神经网络拟合干旱经验频率的方法,避免了事先假定其理论分布形式的不足。在此基础上,借助模拟技术对灌溉工程的效益进行了风险评估;将模糊物元模型引入到灌溉方案优选问题研究中,结合协商定权和熵权理论,实现了灌溉备选方案的优选。这些成果能为研究区的长期规划提供科学的依据。更多还原

【Abstract】 With rapid economy development and population increase, it is more and more important for people to take efficient operations for agricultural drought disaster. We have known that drought disaster can not be avoided and only reduce loss by all means. Traditional methods get inapplicable under serious condition of water deficit. It is necessary to take risk management for drought disaster. How to realize efficient management for drought? This is the hotspot and difficulty in water resources field. Agriculture is basis of national economy in our country. It is very urgent to make research to risk management of agriculture drought thoroughly. This paper establishes the system of risk management for agricultural drought disaster on basis of the research trend in the world, puts forward methods for describe regulations of agricultural drought in time and space based on uncertain and incomplete theory, and suggests risk assessment model for agricultural drought with the technique of data mining. The main results of this paper are as follows:(1) Point out that drought definition relates to two aspects of water demand and water supply, and abnormal feature of drought should be considered. Based on this idea, improved drought definition that is suit to practice is supposed. Suggest extended risk and extended risk of natural disaster that describe conceptions by disaster influence and forecasting liability. On the basis of extended risk, system of risk management for agricultural drought is established.(2) Cloud theory and information diffusion technology are used to take research to danger of agricultural drought. Models with these theory and technology can take advantage of more information, describe greatly non-linear relations between anniversary precipitations, and get better results than other classic methods.(3) Point out it is the most important that drought evaluation for agriculture should correctly reflect the loss caused by the agricultural drought, and establishes the quantitative risk assessment models for agricultural drought that are static model and dynamic model respectively. These models can not only describe the drought extent quantitatively but also reflect the agricultural lose caused by drought. Model of genetic programming is established to describe water production function. Model that is more flexible and intelligent can find the optimal structure automatically by evolutionary algorithm and can avoid the inconvenience of establishing concrete formula.(4) Put forward risk decision model for agricultural drought based on group decision. This model can reduce faults result from the prejudice of individual. Improve traditional particle swarm optimization and apply it into optimal design of agricultural water resources. Suggest carrying capability of agricultural water resources and establish corresponding model.(5) Make research to probability risk of agricultural drought based neural networks optimized by chaos methods, which can avoid the inconvenience of establishing concrete mathematical formulas and the calculation of parameters. Take advantage of simulation technology to realize the benefits calculation of irrigation project. Apply fuzzy matter-element into decision of irrigation schemes, and realize synthetic assessment for irrigation schemes.更多还原