双层多种群PSO在水库群供水优化调度中应用  

Two-layer multi-swarm particle swarm optimal algorithm with application to optimal water supply dispatching of multi-reservoirs

作　　者：卿逸男[1] 丁永生[1,2] 曾献辉[1,2] 郝矿荣[1,2] 

机构地区：[1]东华大学信息科学与技术学院,上海201620 [2]数字化纺织服装技术教育部工程研究中心,上海201620

出　　处：《计算机工程与应用》2013年第5期263-267,共5页Computer Engineering and Applications

基　　金：国家自然科学基金重点项目(No.61134009);国家自然科学基金(No.60975059);教育部高等学校博士学科点专项科研基金(No.20090075110002);上海市优秀学术带头人计划项目(No.11XD1400100);上海领军人才专项资金;上海市科学技术委员会重点基础研究项目(No.11JC1400200);中央高校基本科研业务费专项资金

摘　　要：针对水库群供水优化调度问题,提出了一种带差分进化的双层多种群粒子群算法(DE-TMPSO)。该算法实现粒子群优化算法的群体拓展和双并行运行机制,针对性地提高粒子群算法的全局搜索能力,同时采用不同粒度的多子群并行机制、种群间的双向最优信息流动以及引入差分进化策略也提高了该算法的局部搜索能力,在一定程度上避免了"早熟"现象的发生,具有较好的稳定性,收敛速度也得到了提高。该算法应用于我国南方某流域的水库群供水优化调度问题中,调度结果合理,为求解高维、复杂的水库群供水优化调度提供了新的思路和方法。Aiming at the problem of optimal water supply dispatching for multi-reservoirs, it presents a Two-layer Multi-swarm Particle Swarm Optimal algorithm with Differential Evolution（DE-TMPSO）. The DE-TMPSO realizes the swarm size expan- sion and the dual parallel-running mechanism, so it can purposefully enhance the global search ability. Meanwhile the different granularity in multi sub-swarms parallel mechanism, dual direction optimal information flow between sub-swarms and differen- tial evolution strategy also increase the local search ability. The DE-TMPSO can avoid the premature problem and increase the stability and the convergence rate. The DE-TMPSO is applied to optimal multi-reservoir water supply dispatching of a river in the south China. Results show that the DE-TMPSO is reasonable, and it provides a new approach for multi-dimensional and complicated optimization of multi-reservoir water supply dispatching.

关 键 词：供水优化调度 水库群 多种群粒子群 差分进化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]