水文模型单目标优化和水库群调度多目标决策方法研究

发布时间：2018-06-02 02:45

本文选题：水文模型 + 参数优选　；参考：《华中科技大学》2016年博士论文

【摘要】：水文预报和流域水资源利用及管理是水科学研究中的两大重要内容,前者通过各类水文模型实现,后者主要依靠修建水库枢纽、引调水工程等工程设施和水库调度等非工程措施。其中,流域水库群的联合优化调度是充分发挥其防洪、发电、供水、航运和生态等水资源综合效益,实现经济社会可持续发展的有效工具。此外,水库群系统的调度运行高度依赖可靠的水文预报信息,同时又对水文系统造成影响,使得自然条件下水文现象的演变规律发生改变,给水资源的开发利用带来巨大挑战,两者关系密切。因此,寻求有效的水文模型参数率定方法以提高水文模型的预报精度、探索合理的水库群调度规则方案一直是水科学中面临的两类关键技术难题。本文围绕解决水文预报模型单目标参数优化和水库群调度规则多目标决策这两个参数类优化问题展开研究,主要工作和创新性成果如下：(1)为解决非线性水文预报模型参数全局优化率定问题,开发了一种混合遗传布谷鸟算法(GACS),4个复杂约束优化测试问题的仿真实验检验了GACS算法的有效性。GACS算法通过实数编码个体方式提高算法的执行效率,采用自适应交叉变异机制和精英保留策略强化算法的寻优能力和收敛性能；利用基于个体适应度和约束违反项两属性比较的锦标赛选择算子处理约束条件,巧妙地回避了罚函数法中罚参数的优选难题；引入标准布谷鸟算法的Levy flight搜索机制,扩大种群搜索范围和多样性,进一步提高算法的全局搜索能力。(2)建立了一种改进的非线性变指数参数马斯京根模型-VEP-NLMM-L,成功应用GACS算法实现了对该模型的参数优化率定,并通过发生在不同流域、不同类型的典型洪水计算实例(算例1-3)验证了VEP-NLMM-L模型的有效性和适用性。针对传统线性和非线性马斯京根模型存在考虑因素不全、预报精度低及适应性弱等缺点,VEP-NLMM-L模型加入如下改进：①引入指数参数β建立非线性槽蓄曲线方程,②加入旁侧入流系数参数x考虑沿河道旁侧入流的影响,③划分不同入流级别(i=1,2,…,L)并采用不同指数参数值βt表征河道洪水传播的非恒定流特性,④模型包含(2L+2)个参数、(L-2)个模型结构约束。(3)构建了NSGA-Ⅱ—EABODE多目标决策方法框架。NSGA-Ⅱ算法解决多目标优化问题以生成有限备选决策方案集,SEABODE多属性决策方法完成对备选决策方案集的进一步评价、排序和筛选问题；SEABODE法是一种基于k阶p级有效或[k,p]-Pareto最优概念的决策方案逐次淘汰方法,它克服了传统聚合类多属性决策方法(AHP、TOPS、Vague集、灰色关联分析等)存在的一些问题；数值实例演示了SEABODE法进行多属性决策的过程,说明了该方法的有效性。(4)应用NSGA-Ⅱ—SEABODE法进行了嘉陵江枯水年水库群供水调度规则优化提取及筛选决策研究。首先,设计了一种由单个水库改进限制供水规则和并联水库群共同供水任务分配水规则组成的水库群供水调度规则；然后,建立了由水库供水风险指标体系和反映用水户缺水影响的缺水指标共同构成的水库(群)供水调度规则评价指标体系：其次,以水库群供水系统总供水缺水率和单个水库时段供水任务的破坏深度最小为目标函数,建立了嘉陵江水库群供水调度多目标优化模型：最后,针对枯水典型年,利用NSGA-Ⅱ算法提取了水库群非劣供水调度规则方案集,并采用SEABODE法依据上述评价指标体系筛选出各水库偏好的调度规则方案。
[Abstract]:Hydrological forecast and water resource utilization and management are two important contents in water science research. The former is realized through various hydrological models, the latter mainly relies on the construction of reservoir hubs, water diversion projects and other non engineering measures, such as engineering facilities and reservoir scheduling. The comprehensive benefits of water resources, such as electricity, water supply, shipping and ecology, are effective tools for the sustainable development of the economy and society. In addition, the operation of the reservoir system is highly dependent on the reliable hydrological forecast information, at the same time, it also affects the hydrological system, making the evolution of the hydrological phenomena under natural conditions changing and the development of water resources. With great challenges, the two are closely related. Therefore, the search for effective hydrological model parameter rate determination method to improve the forecasting accuracy of hydrological model and to explore the rational scheme of reservoir group scheduling have been two key technical problems in water science. The main work and innovative results are as follows: (1) a hybrid genetic cuckoo algorithm (GACS) is developed to solve the global optimization problem of nonlinear hydrological forecast model, and the simulation experiment of 4 complex constrained optimization tests is made to test the GACS algorithm. The effective.GACS algorithm improves the execution efficiency of the algorithm through the real number coding individual, and uses the adaptive cross mutation mechanism and the elite reservation strategy to strengthen the optimization ability and convergence performance of the algorithm. The tournament selection operator based on the comparison of individual fitness and constraint violation two attributes is used to deal with the constraints, and the penalty is skilfully avoided. The Levy flight search mechanism of standard cuckoo algorithm is introduced to expand the range and diversity of the population search and further improve the global search capability of the algorithm. (2) an improved nonlinear variable exponent parameter -VEP-NLMM-L is established, and the GACS algorithm is successfully applied to the model. The parameter optimization rate is determined, and the validity and applicability of the VEP-NLMM-L model are verified by the example of typical flood calculation in different basins and different types (example 1-3). In view of the shortcomings of the traditional linear and nonlinear Ma shjing root model, such as incomplete factors, low prediction precision and weak adaptability, the VEP-NLMM-L model is added to the following changes. The following: (1) introducing the exponential parameter beta to establish the nonlinear trough storage curve equation; (2) adding the side inflow coefficient parameter x to consider the influence of the inflow along the side of the channel; (3) dividing the different inflow levels (i=1,2,... L) and using different exponential parameter values of beta T to characterize the unsteady flow characteristics of river flood propagation, (3) (L-2) model structure constraints. (3) a NSGA- II EABODE multi-objective decision method framework.NSGA- II algorithm is used to solve multi-objective optimization questions to generate a finite alternative decision scheme set, SEABODE multiple attribute decision making The method completes the further evaluation, sorting and screening of the alternative decision sets. The SEABODE method is a successive elimination method based on the k order p class effective or [k, p]-Pareto optimal concept, and it overcomes some problems existing in the traditional aggregator multi attribute decision-making method (AHP, TOPS, Vague set, grey relational analysis, etc.); The example demonstrates the process of multi attribute decision making by SEABODE method and illustrates the effectiveness of the method. (4) the optimization extraction and selection of water supply scheduling rules of the Jialing River in the dry year of the Jialing River are studied by using the NSGA- II SEABODE method. The regulation of water supply dispatching in the reservoir group consists of water rules. Then, the index system of reservoir water supply risk index system and the water shortage index which reflect the water shortage effect of the water users are established. Secondly, the water shortage rate of the total water supply of the reservoir water supply system and the water supply task of the single reservoir period are given. The minimum damage depth is the objective function, and the multi-objective optimization model of water supply scheduling for the Jialing River reservoir group is established. Finally, according to the typical year of the dry water, the NSGA- II algorithm is used to extract the scheme set of the non inferior water supply scheduling rule of the reservoir group, and the scheduling rules of each reservoir preference are selected by the SEABODE method.
【学位授予单位】：华中科技大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TV697.12

【相似文献】