Application of dynamic programming to control khuzestan water resources system
Application of dynamic programming to control khuzestan water resources system
复制标题
DOI：
10.1016/0005-1098(77)90055-3
复制
发表时间：
1977-05
期刊：
影响因子：
--
通讯作者：
M. Jamshidi;M. Heidari
中科院分区：
文献类型：
--
作者：
M. Jamshidi;M. Heidari
采用离散微分动态规划（DDDP）方法，对伊朗胡泽斯坦省的水资源系统进行了优化调度。该技术利用每个状态变量的初始标称状态轨迹，并在轨迹周围形成走廊。这些走廊表示整个可行域的一组子域。从这样一组名义状态轨迹开始，在围绕它们形成的走廊内寻求目标函数的改进。这导致了一组新的标称轨迹，在此基础上可以寻求更多的改进。由于优化被限制在一组子域，在内存和计算机时间上实现了相当大的节省比传统的动态规划。库兹斯坦水资源系统位于伊朗西南部，由两条河流、三座水库、三座水电站和三个可灌溉区组成。用于分析的数据和成本效益函数从历史记录或类似研究中获得。
An approximate optimization technique based on discrete dynamic programming called discrete differential dynamic programming (DDDP), is employed to obtain the near optimal operation policies of a water resources system in the Khuzestan Province of Iran. The technique makes use of an initial nominal state trajectory for each state variable, and forms corridors around the trajectories. These corridors represent a set of subdomains of the entire feasible domain. Starting with such a set of nominal state trajectories, improvements in objective function are sought within the corridors formed around them. This leads to a set of new nominal trajectories upon which more improvements may be sought. Since optimization is confined to a set of subdomains, considerable savings in memory and computer time are achieved over that of conventional dynamic programming. The Kuzestan water resources system considered in this study is located in southwest Iran, and consists of two rivers, three reservoirs, three hydropower plants, and three irrigable areas. Data and cost benefit functions for the analysis were obtained either from the historical records or from similar studies.