【作者】 侯召成；

【导师】 王本德；

【作者基本信息】 大连理工大学 ， 水文学及水资源， 2004， 博士

【摘要】 作为主要防洪工程措施之一，水库对保证国民经济稳定增长，保障人民生命财产安全和社会的安定发挥了重要作用。合理的水库防洪预报调度能够有效的解决或缓解水库防洪调度中存在的主要矛盾，有利于防洪与兴利效益的有机结合。 在总结水库防洪调度的任务、分类及主要特点的基础上，本文首先综述了模糊集理论及风险分析理论在水库防洪系统中的发展和应用，然后概括了水库防洪调度的发展趋势。鉴于水库防洪预报调度所依据信息的复杂性，本文主要对水库防洪预报调度中存在的模糊现象和风险分析问题进行研究，研究内容和成果概括如下： (1)在简要回顾模糊聚类理论及应用研究进展和介绍模糊聚类迭代模型的基础上，提出了基于类间相关系数的聚类有效性公式，随后的实例分析和水库洪水聚类分析验证了所提有效性公式的有效性和实用性。 (2)在王本德教授的水库防洪预报研究成果的基础上，分析了水库防洪调度设计的存在问题，提出了水库防洪分类预报调度设计方法。基于实测水文资料，结合洪水过程的成因分析和聚类分析，对洪水过程进行综合分析；在此基础上，结合水库的调度经验，考虑天气系统、雨强等因素对洪水调度的影响，通过分别调节不同天气系统成因的预报设计洪水过程，以期建立满意的水库防洪分类预报调度规则。随后以碧流河水库为例，详细阐述了水库分类防洪预报调度规则的确定过程，初步验证了水库防洪分类预报调度设计方法的可行性。 (3)针对水库防洪调度的群决策性，在陈守煜教授提出的多目标半结构性决策理论的基础上，首先引入数学转换将决策者给出的模糊判断矩阵转换为满足完全一致性的模糊判断矩阵，然后通过简单加权集结方法将群体决策集结为单人决策，试图将陈守煜提出的理论推广到群决策环境下，拓宽其应用范围。 (4)在总结现有水库防洪系统风险分析方法存在问题和模糊风险分析研究现状的基础上，引入可能性一概率风险的概念来进行水库防洪系统风险分析，并以白龟山水库降雨预报的精度分析为例说明该方法的具体应用。可能性—概率风险概念简单，数学逻辑清晰，计算简便，特别适合于因历史数据少而形成的小样本问题。 (5)在王本德教授提出的风险率计算方法的基础上，给出了设计允许极限误差的定义以及水库和下游防洪风险度的定量化计算公式，并提出了水库防洪预报调度风险分析的基本思路和设计允许极限误差的计算原理与方法。随后以碧流河水库为例，主要就不同极限指标控制下的设计允许极限误差及不同水库运行状态与不同误差组合下的水库防洪风险度进行了计算，并进行了结果分析。最后对全文进行了总结，并对有待进一步研究的问题进行了展望。更多还原

【Abstract】 As one of the main structural flood control measures, reservoir plays an important role in national economic development and social stability. Rational reservoir forecasted flood operations may effectively handle or mitigate the main contradictions existed in reservoir flood operations, and may favor the effective integration of flood control and beneficial functions of reservoirs.Based on a summary of the task, classification and main characteristics of reservoir flood operations, the applications of fuzzy set and risk analysis theory to reservoir flood operations are first generalized in this paper, and then the tendency of reservoir flood operations is pointed out. Aiming at the information complexity, the fuzzy phenomena and risk analysis problems are mainly investigated, and the research results are listed as follows.(1) With a brief review of fuzzy clustering theory development and an introduction of fuzzy clustering iterative algorithm, correlated coefficient-based cluster validity functional is presented, and then the followed experimental analysis and flood clustering analysis testify the efficacy and applicability of the proposed functional.(2) On the basis of the analysis of the existing problems in the design for reservoir flood operations, the classified design method for reservoir forecasted flood operations is proposed. Founded upon the observed hydrological data, floods are analyzed in an integral way. Taken the weather system and rainfall intension into consideration, the classified dispatching rules are determined for reservoir forecasted flood operations through regulating the forecasted floods with different weather systems. Taken Biliuhe reservoir as an example, the flow for determining the classified dispatching rules is systematically illustrated, and the applicability of the proposed method is testified.(3) Aiming at the group decision making characteristic of reservoir flood operations, and on the basis of multiobjective semi-structural decision-making theory proposed by Professor Chen Shouyu, a mathematical transformation is introduced to convert thefuzzy judgment matrices provided by the decision makers into those with complete consistency, and then a simple weighted method is applied to integrate group decision makings into decision makings with single decision maker, and tries to extend the decision making theory proposed by Chen into group decision making environment.(4) With a summary of the existing problems in reservoir flood operation risk analysis and the current research of fuzzy risk analysis, possibility-probability risk analysis method is introduced for reservoir flood operation risk analysis, and the precision analysis of rainfall forecast for Baiguishan reservoir is provided as an example to illustrate the application of the presented method. With a feature of simple concept, clear mathematical logic and easy computation, possibility-probability risk analysis method is particularly suitable for small sample problems with scant history data.(5) With the method proposed by Professor Wang Bende for the computation of the limiting risk rate, the concept of the allowable limiting forecast errors in design manner is given, and the quantified formulae for the risk degrees of both reservoir and downstream area are yielded. Meanwhile, the basic flow for reservoir flood operation risk analysis and the principle and method for the allowable limiting forecast errors in design manner are presented. Taken Biliuhe reservoir as an example, flood operation risk of both reservoir and downstream area is calculated, and the computation results are discussed, too.Finally, the concluding part is given, and some problems to be further investigated are discussed.更多还原