梯级水库发电优化调度搜索空间缩减法及其应用

Search space reduction method and its application to hydroelectric operation of multi-reservoir systems

梯级水库联合调度是对区域水资源进行直接调控的最有效措施之一,相比单库调度,可显著增加社会、经济和生态效益。针对智能算法求解梯级水库优化调度模型搜索效率较低、寻优结果不稳定等缺陷,本文基于发电调度模型中水电站最小出力以及下泄流量约束,提出了搜索空间缩减法。将其与智能算法耦合,利用优化的搜索空间产生高质量的初始种群,同时使算法在更小的搜索空间内寻优,进而提升算法的搜索效率。以典型入库流量下的某梯级水库发电优化调度为例,选用布谷鸟算法进行优化计算,对比了传统搜索空间与优化的搜索空间对算法搜索效率的影响。实例研究表明:缩小可行空间方法可进一步提升智能算法的收敛性以及求解精度,是改善梯级水库调度模型求解效率的一种实用、有效方法。

Joint operation of multi-reservoir systems is one of the most effective measures for direct control of regional water resources and it significantly increases the social, economic and ecological benefits compared with single reservoir operation. This paper describes a search space reduction method （SSRM） for solving hydroelectric optimal operation models of multi-reservoir systems, using the minimum output of a hydro plant and its discharge constraints to reduce the search space and overcome the shortcomings of the meta-heuristics in lower search efficiency and instable results. SSRM, when coupled with the meta-heuristics, can produce a high-quality initial population, and its search process is limited to an optimal search space that is closer to the feasible search domain. Thus, such reduction makes it easier to find the globally-optimized solution. We have adopted an effective algorithm, namely the cuckoo search （CS）, in the optimization, and compared the traditional and reduced search spaces for a multi-reservoir system with given historical monthly inflows. A case study shows that SSRM, a simple and effective method for practical use, improves convergence and solution accuracy of the meta-heuristics.