跳跃海豚群算法  

Jumping dolphin swarm algorithm

作　　者：王艳娇 史新梦 WANG Yan-jiao;SHI Xin-meng(School of Information Engineering, Northeast Electric Power University, Jilin Jilin 132000, China)

机构地区：[1]东北电力大学电气工程学院

出　　处：《控制理论与应用》2019年第10期1755-1767,共13页Control Theory & Applications

基　　金：国家自然科学基金项目(61501107);吉林省教育厅“十三五”科学技术研究(吉教科合字[2016]第95号);吉林市科技创新发展计划项目(201750219)资助~~

摘　　要：针对海豚群算法收敛速度慢、容易陷入局部最优等缺陷,提出一种跳跃海豚群算法.增加跳跃步骤,保留更多优秀解,加快收敛;提出声波长度随迭代自适应变化的策略以及含有变异因子的动态位置更新策略,满足算法不同时期对进化的需求;并加入早熟收敛机制,降低陷入局部最优的机率.最后,分析了参数对算法性能的影响,并与四种算法进行对比,实验表明其在收敛速度、收敛精度以及鲁棒性上优势明显.Aiming at the shortages of dolphin swarm algorithm (DSA) with slow convergence speed and easy to fall into local optimum, a jumping dolphin swarm algorithm (JDSA) was proposed. By adding jumping steps, more excellent solutions are retained to accelerate convergence speed. A strategy of making the sound wave-length adaptive to the change of iteration and a strategy of renewing the dynamic position containing the variation factor are proposed to meet the requirements of different periods of the algorithm. The premature convergence mechanism is added to reduce the probability of falling into local optimum. Finally, the influence of parameters on the performance of the algorithm is analyzed and compared with four algorithms. Experiments show that the algorithm has obvious advantages in convergence speed, convergence accuracy and robustness.

关 键 词：海豚群算法 函数优化 长度自适应策略 跳跃步骤 早熟收敛机制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]