求解水库优化调度问题的动态规划-遗传算法  

Optimization of reservoir operation by using dynamic programming-genetic algorithm

作　　者：刘攀[1] 郭生练[1] 雒征[1] 刘心愿[1] 

机构地区：[1]武汉大学水资源与水电工程科学国家重点实验室,湖北武汉430072

出　　处：《武汉大学学报（工学版）》2007年第5期1-6,共6页Engineering Journal of Wuhan University

基　　金：国家自然科学基金(编号:50609017);湖北省自然科学基金(编号:2006ABA229)

摘　　要：综合动态规划和遗传算法的优点,基于贝尔曼最优化原理将水库优化调度按阶段划分为若干多目标决策子问题,各子问题采用混合编码的多目标遗传算法求解,从而提出了一种求解水库优化调度问题的动态规划-遗传算法.该算法不仅在时间上通过划分阶段降维,而且采用遗传算法克服离散状态空间组合所产生的维数灾问题.从理论上论证了动态规划-遗传算法的全局收敛性,分析得出该算法的效率一般高于遗传算法;并经数值试验表明:在计算时段数较多时,动态规划-遗传算法显著优于遗传算法.因此,提出的动态规划-遗传算法为求解水库优化调度问题提供了新的思路与途径.Based on the Bellman＇ s optimality principle, a dynamic programming-genetic algorithm （DPGA） is proposed to optimize the reservoir operation. The reservoir optimal operation is partitioned into multi-stage problems based on the optimality principle, which can be optimized by using multi-objective GA respectively. The dimensions of both temporal and spatial could be reduced greatly; and the global convergence and efficiency of the DP-GA are analyzed theoretically. A Monte Carlo experiment is used to simulate and evaluate the proposed algorithm with the Qingjiang River cascade hydropower stations as case study. It is indicated that the DP-GA is valuable for the reservoir optimal operation, especially for the optimization of reservoir operation with many periods.

关 键 词：水库调度 遗传算法 最优化原理 动态规划-遗传算法 

分 类 号：TV697.1[水利工程—水利水电工程]