摘要
为增强动态规划算法对梯级水库群联合优化调度问题的求解能力,基于多层嵌套的思想,提出了多层嵌套多维动态规划算法。通过与传统多维动态规划算法的对比分析,发现该方法能在内存占用量以及程序复杂度2方面有效缓解多维动态规划算法应用于梯级水库群联合优化调度时的维数灾问题。针对该方法在运行时间上的缺陷,引入并行处理技术,利用状态离散点间的计算独立性构建了多层嵌套多维动态规划并行算法。以李仙江流域三库梯级为背景进行了实例研究,并从运行时间、并行加速比、并行效率以及CPU核数等方面对该并行算法进行了详细的性能分析。结果表明在现有的计算条件下该并行算法能有效提高计算效率和缓解多维动态规划维数灾问题。
In order to enhance the dynamic programming algorithm＇s ability to solve the joint operation optimization problem of cas-cade reservoirs ,the multilayer nested dynamic programming algorithm is proposed in this paper based on the thought of multi-layer nested structure .Through a comparison with the conventional multi-dimensional dynamic programming algorithm ,it is found that this method can effectively alleviate the dimension disaster problem of multi-dimensional dynamic programming algorithm in the ap-plication of cascade reservoirs operation optimization in terms of programming complexity and memory space occupation .Targeted at the defect on run-time ,this paper introduces the parallel processing technology and builds the multilayer nested parallel multi-dimen-sional dynamic programming algorithm by using the independence of calculation among the discrete points .Li Xianjiang River cascade reservoirs is used as the study instance in this paper ,and there is a detailed performance analysis of this parallel algorithm on run-time ,parallel speedup ,parallel efficiency and the number of CPU cores .Results show that this parallel algorithm can effectively im-prove the computation efficiency and alleviate the dimension disaster of multi-dimensional dynamic programming under the condition of the existing computer hardware .
出处
《中国农村水利水电》
北大核心
2014年第9期70-75,共6页
China Rural Water and Hydropower
基金
国家自然科学基金资助项目(51279062)
中央高校基本科研业务专项资金项目(13XS23
13XS22
13QN22)
关键词
梯级水库
多维动态规划
并行算法
优化调度
李向江流域
cascade reservoirs
multi-dimensional dynamic programming
parallel algorithm
operation optimization
Lixianjiang Basin
