摘要
将多维动态规划算法与并行处理技术相结合,是增强该算法对梯级水库群联合优化调度问题求解能力和缓解其"维数灾"问题的有效措施。基于.NET4的并行拓展库(Parallel Extensions),构建了时段间、时段内离散组合间以及两者混合模式下的并行多维动态规划算法,以我国某流域两库梯级为研究背景进行了实例计算,并从运行时间、并行加速比、并行效率以及内存占用等方面对这三种并行模式下的并行多维动态规划算法进行了详细的对比分析,结果表明三种并行模式都能有效缩短程序运行时间,从而缓解多维动态规划的维数灾问题,但相对而言时段内离散组合间并行模式最优,混合并行模式次之,时段间并行模式最差。
In order to enhance the ability and alleviate the curse of dimensionality of multi-dimensional dynamic programming algorithm in solving the joint operation optimization problem of cascade reservoirs,We take effective measures to combine this algorithm with the parallel processing technology.This paper has proposed three parallel multi-dimensional dynamic programming algorithms based on NET 4Parallel Extensions by using the stages parallel mode,state combinations parallel mode and hybrid parallel mode.A certain cascade reservoir in China is used as a case study in this paper,and a detailed contrastive analysis of the three parallel multidimensional dynamic programming algorithms on run-time,parallel speedup,parallel efficiency and memory usage has been carried out based on the calculation results.Results show that the parallel algorithms can effectively shorten the program run-time and alleviate the curse of dimensionality to multi-dimensional dynamic programming algorithm,but relatively speaking,the state combinations parallel mode is optimal,the hybrid parallel is suboptimal and the stages parallel mode is the worst.
出处
《中国农村水利水电》
北大核心
2015年第3期168-173,176,共7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)
关键词
梯级水库
水库调度
多维动态规划
维数灾
并行算法
并行模式
cascade reservoirs
reservoir operation
multi-dimensional dynamic programming
curse of dimensionality
parallel algorithm
parallel mode
