Particle Swarm Optimization Algorithm Based on Combing Example Learning and Opposition Learning
- 分享到:
摘要:
为了提高粒子群优化算法(Particle swarm optimization,PSO)的优化效率,降低其陷入局部最优的概率,提出了一种融合榜样学习和反向学习的PSO算法(PSO based on combing Example learning and Opposition learning,EOPSO).首先,对粒子群中的非最优粒子采用新颖的榜样学习机制更新,以便提高全局搜索能力,避免算法陷入局部最优;其次,对粒子群中最优粒子采用反向学习混合机制更新,提升该粒子的搜索能力,进一步避免算法陷入局部最优;最后,对粒子群中的最优粒子还采用了自身变异机制更新,有利于搜索前期的全局搜索和后期的快速收敛.在15个不同维度的基准函数上进行了仿真实验,实验结果表明,与最先进的PSO改进算法ELPSO、SRPSO、LFPSO、HCLPSO相比,EOPSO优化性能更好.
In order to improve the optimization efficiency of the particle swarm optimization algorithm and prevent the al-gorithm from trapping into the local optima. Based on combing Example learning and Opposition learning （EOPSO）. This pa-per proposes a PSO Firstly, all non-optimal particles in the particle swarm are updated by a novel example learning mechanism to improve their search ability and to prevent the algorithm from trapping into the local optima. Secondly, the optimal particle is updated by a hybrid opposition learning way to improve its search ability and further avoid the algorithm^ trapping into the local optima. Finally, a self-mutation mechanism is also adopted to update the optimal particle to increase the population diver-sity. In addition, the self-mutation mechanism adopts an adaptive mutation rate to provide the good global search ability at the early search phase and accelerate the convergence speed at the late search phase in the algorithm process. The simulation experi-ments are made on 15 benchmark functions with different dimensions. The experiment results show that, compared with the state-of-the-art PSO variants such as ELPSO, SRPSO, LFPSO and HCLPSO, EOPSO obtains better optimization perform-ance.
作者:
张新明 王霞 涂强 康强
机构地区:
betway官方app 计算机与信息工程学院 betway官方app 计算智能与数据挖掘河南省高校工程技术研究中心
出处:
《betway官方app 学报：自然科学版》 CAS 北大核心 2017年第6期91-99,共9页
基金:
河南省重点科技攻关项目(132102110209) 河南省基础与前沿技术研究计划项目(142300410295)
关键词:
智能优化算法 粒子群优化算法 榜样学习 反向学习
intelligent optimization algorithm particle swarm optimization algorithm example learning opposition learning
分类号:
TP181 [自动化与计算机技术—控制理论与控制工程]
