摘要
以差分进化算法(DE)为基本框架,结合混沌算法(CA)和蛙跳算法(SLFA)各自局部搜索优势以及多核并行计算技术(PC),提出一种新的并行混合差分进化算法(PHDE),即将DE与CA、SLFA进行有机融合,分别对精英个体进行混沌局部搜索和对较差个体进行蛙跳局部更新,且差分进化运算、混沌局部搜索和蛙跳局部更新均采用PC,以有效缩短计算时间。PHDE具有三点优势:一是保留了DE简单易行、收敛迅速的特点;二是继承了CA、SLFA的遍历性,能够避免早熟收敛现象;三是通过合理的并行模式,有效降低了计算时间。典型测试函数表明了PHDE的可行性、高效性和鲁棒性。实例研究表明,PHDE具有较好的优化性能和计算效率,为高效求解水库群优化调度问题提供了一种可行途径。
This paper presents a new parallel hybrid differential evolution （PHDE） algorithm for optimal scheduling of cascade reservoirs by combining a chaos algorithm （CA） and a shuffled leapfrog algorithm （SLFA） for their advantages in local search with parallel computation （PC） technique and using a basic framework of difference evolution （DE）. PHDE is an organic fusion of DE, CA, SLFA and PC, i.e. conducting chaotic local search for better individuals and leapfrog local search for worse individuals, and its computation of DE, CA and SLFA is combined with PC so as to reduce running time. Therefore, this algorithm has three advantages： it reserves the simplicity and fast convergence of DE; it inherits the ergodicity from CA and SLFA so as to avoid premature convergence; it has a relative shorter computing time. The testing results of typical functions verified its better feasibility, efficiency and robustness. Case studies showed that it can achieve better optimization performance, thus offering an efficient tool for solving optimal operation model of cascade reservoirs.
出处
《水力发电学报》
EI
CSCD
北大核心
2017年第6期57-68,共12页
Journal of Hydroelectric Engineering
基金
国家重点研发计划重点专项项目(2016YFC0402202)
关键词
梯级水库群
优化调度
差分进化
混沌算法
蛙跳算法
并行计算
cascade reservoirs
optimal scheduling
differential evolution algorithm
chaos algorithm
shuffled leapfrog algorithm
parallel computation
