›› 2018, Vol. 31 ›› Issue (1): 20-.

• 论文 • 上一篇    下一篇

并行人工蜂群算法研究

 梅凯, 火久元, 常扣扣   

  1. 兰州交通大学 电子与信息工程学院
  • 出版日期:2018-01-15 发布日期:2018-01-11
  • 作者简介:梅凯(1989-),男,硕士研究生。研究方向:智能计算、并行计算等。 火久元(1978-),男,博士,副教授。研究方向:智能计算、并行计算、数据挖掘等。
  • 基金资助:

    国家自然科学基金(61462058);甘肃省自然科学研究基金计划(1606RJZA004);2016年赛尔网络下一代互联网技术创新项目(NGII20160111)

A Parallel Approach for Artificial Bee Colony Algorithm

MEI Kai,HUO Jiuyuan,CHANG Koukou   

  1. School of Electronic and Information Engineering,Lanzhou Jiaotong University
  • Online:2018-01-15 Published:2018-01-11

摘要:

针对人工蜂群算法在处理高维度问题时收敛速度慢的问题,利用OpenMP多线程技术和规约机制,并根据已改进的观察蜂来选择雇佣蜂的方式,提出了基于OpenMP的并行人工蜂群算法(PCABC)。仿真实验分别在问题维度为100和200下进行来评估算法性能,在4个逻辑处理器环境下,基于静态调度的并行人工蜂群算法的加速比最高可以达到3.95,效率可达98.65%。实验结果表明,PCABC并行人工蜂群算法在处理高维度复杂函数时,收敛速度和算法运行时间都有较大的提升。

关键词: 人工蜂群算法, 人工蜂群算法改进, 群体智能, 并行化, OpenMP并行处理

Abstract:

Aiming at the slow convergence speed of artificial bee colony algorithm in dealing with high dimensional problems, this paper uses OpenMP multi-threading technology and regulation mechanism, and according to the improved way onlooker bees choose employed bees, a parallel artificial bee colony algorithm(PCABC) based on OpenMP is proposed. Simulation experiments are performed to evaluate the performance of the algorithm under three different types of cyclic parallel scheduling types of OpenMP. In the 4 core processor environment, the speedup of parallel artificial bee colony algorithm based on static scheduling can reach to 3.95, the efficiency can reach to 98.65%.The experimental results show that the PCABC parallel artificial bee colony algorithm has higher lifting speed and running time when dealing with high dimensional complex functions.

Key words: artificial bee colony algorithm;improved artificial bee colony algorithm;swarm intelligence;parallelism, OpenMP parallel processing