Journal of Frontiers of Computer Science and Technology ›› 2016, Vol. 10 ›› Issue (11): 1587-1600.DOI: 10.3778/j.issn.1673-9418.1510034

Previous Articles     Next Articles

Fruit Fly Optimization Algorithm by Combining Strategies of Swarm Collaboration and Harmony Search

LIU Le+   

  1. School of Management, University of Jinan, Jinan 250002, China
  • Online:2016-11-01 Published:2016-11-04


刘  乐+   

  1. 济南大学 管理学院,济南 250002

Abstract: To deal with the drawbacks of trapping in local optimal solutions easily and low convergence accuracy of the standard fruit fly optimization (FFO) algorithm, this paper proposes a novel fruit fly optimization algorithm by combining the swarm collaboration (SC) and harmony search (HS) strategies, named as FFO-SC+HS. In each iteration of FFO-SC+HS, the food source location of each fruit fly is generated based on a single dimension that is randomly    determined and dynamic search radius, and two candidate location vectors derived from the reconstructed location set by SC strategy are considered during the update process of fruit fly swarm location. Further, one candidate location is the best one of the reconstructed location set, and the other one is obtained by means of HS strategy. Extensive computational experiments and comparison analysis are conducted upon 10 benchmark functions to validate the effectiveness of FFO-SC+HS. As demonstrated in the results, FFO-SC+HS outperforms other 4 reported FFO algorithms in terms of solution quality and convergence efficiency. Moreover, it is found that different combinations of three main parameters have a significant impact on optimization performances of FFO-SC+HS, and both of the SC and HS strategies are     indispensable.

Key words: fruit fly optimization, swarm collaboration, harmony search, function optimization

摘要: 为了改善标准果蝇优化(fruit fly optimization,FFO)算法易陷入局部极优,收敛精度不高的不足,提出了一种结合群体协同(swarm collaboration,SC)与和声搜索(harmony search,HS)策略的新型果蝇优化算法FFO-SC+HS。该算法基于随机确定的单一维度和动态搜索半径得到果蝇个体的食物源位置,并在种群中心位置的逐代更新环节新增了两个可供选择的备选位置。两备选位置均出自按群体协同策略重构后的位置集合,其一为重构后位置向量集合中的最佳位置,另一则为借助和声搜索策略得到的新位置向量。为验证所设计算法的有效性,在10种测试函数上进行了大量的计算实验与性能对比分析,结果表明FFO-SC+HS在求解质量、收敛能力上优于其他4种已报道的FFO算法,并发现3个主要参数的不同取值组合对其优化性能具有显著影响,所采取的SC与HS策略缺一不可。

关键词: 果蝇优化, 群体协同, 和声搜索, 函数优化