节点文献

水库优化调度函数研究

Research on Optimal Operation Function of Reservoir

【作者】 左吉昌

【导师】 李承军;

【作者基本信息】 华中科技大学 , 系统分析与集成, 2007, 硕士

【摘要】 水库/水库群优化调度的目的是制定优化的水库调度规则,以期望获得更好的效益。它有多种方法,如显随机优化和隐随机优化。显随机优化方法的主要困难是概率转移矩阵的确定和求解过程中的“维数灾”问题。隐随机优化方法主要包括两个步骤:(1)利用历史径流序列资料,用确定性优化算法为水库群寻求优化运行方式;(2)利用确定性优化计算得出的水库运行结果,选择与调度决策相关的运行要素,通过回归分析建立回归方程作为调度规则函数,以指导水库/水库群的联合运行。本文首先介绍了水库优化调度的背景、发展历史、一般原理和方法,然后着重介绍隐随机算法的模型和方法,以及本文中会应用到的逐次优化算法和支持向量机(SVM),并通过实例计算和不同方法得到的调度函数计算结果来说明该方法的应用过程和优点,最后总结方法的优缺点和一些思考、建议。本文重点在于:(1)针对“维数灾”问题,本文采用逐次优化算法,并结合变步长的双向搜索法。该算法能够收敛,得到全局最优解,并且计算时间快。(2)针对水库调度函数的复杂、非线性、以及水文资料的有限性,尝试用支持向量机技术建立水库优化调度函数。并通过实例以及不同方法的比较,证明SVM方法的调度函数具有更好的性能。(3)基于以上两点,本文同时采用信息论的方法进一步分析了水库调度难题的本质,作出一定的探讨和建议,但是水库调度问题本身的复杂性和不确定性,也决定了这一分析方法的难度,长路漫漫,有待进一步探索。

【Abstract】 The optimal operation of reservoir/multireservoir is to develop rules for managing reservoir/multireservoir for more benefits. There are a variety of methods, such as stochastic optimal method and implicit stochastic optimal method. The main difficulty of stochastic optimal method is how to establish the transfer probability matrix and escape the“curse of dimensionality”in the course of solving. Two steps included in implicit stochastic optimal method: (1) Using the deterministic optimization of reservoir operation algorithm and finding out a optimal operation mode with the historical information of runoff. (2) Based on the results of the first step, choose the decision-making relate to the operation and establish the regression equations as a function of scheduling rules through regression analysis of the regression equation, which guides the operation of reservoir/multireservoir.This paper introduces the optimal reservoir management background, history, general principles and methods and expatiates on the Progressive Optimization Algorithm and Support Vector Machine (SVM), which was showed on how to use it and proved more available compared with other methods. There are some thoughts and suggestions at the last.This paper will focus on:(1) In order to solve the“curse of dimensionality”,the Progressive Optimization Algorithm and Variable Step-Size Two-way Search Algorithm are used. They are convergent and get global optimal solution fleetly.(2) According to the complexity and non-linearity of the reservoir operation function and the finiteness of the hydrographic information,an optimization reservoir operation function is established using SVM technology. Compared with different methods and instances, it has been confirmed having better performance than others.(3) Based on two points above, we also discuss the essential difficulty of the optimal operation of reservoir by the Information Theory and give some annotations and suggestions. But the complexity and uncertainty of this difficulty induce a long, difficult and hazardous further exploration, which we will work hard on.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络