基于单增量和全局维度学习策略的萤火虫算法  

Firefly Algorithm Based on Single Increment and Global Dimension Learning Strategy

作　　者：葛婉贞 赵新超[1] 李子旭 吴凌宇 GE Wanzhen;ZHAO Xinchao;LI Zixu;WU Lingyu(School of Science,Beijing University of Posts and Telecommunications,Beijing 100876,China)

机构地区：[1]北京邮电大学理学院,北京100876

出　　处：《河北师范大学学报（自然科学版）》2023年第2期121-129,共9页Journal of Hebei Normal University：Natural Science

基　　金：国家自然科学基金(61973042);北京市自然科学基金(1202020)。

摘　　要：萤火虫算法的搜索过程较依赖于最优萤火虫,而最优萤火虫并不进行有导向的寻优移动,算法易陷入局部最优.为此,提出了一种基于单增量和全局维度学习策略的萤火虫算法.在萤火虫个体移动时,该算法并不叠加萤火虫个体的当前位置,而是将累加的位置增量作为新的搜索方向,用于更新萤火虫的位置.该算法大大降低了萤火虫当前位置对搜索过程的影响,有利于算法更快的跳出当前局部最优,进行更大范围的寻优;其次,对最优萤火虫进行一定次数的单维度学习,将学习后的萤火虫引导种群进化.在基准测试函数上的实验结果表明,该算法优于其他几种改进的群智能优化算法,具有良好的跳出局部最优的能力.The firefly algorithm′s search process is more reliant on the optimal firefly, which does not move in a guided search, making it easy for the algorithm to fall into the local optimum.For which this paper proposes a firefly algorithm based on single incremental and global dimensional learning strategy.When fireflies move individually, the algorithm we proposed does not superimpose their current positions, but instead uses the accumulated position increments as a new search direction to update their positions.This algorithm we proposed greatly reduces the influence of the current positions of fireflies on the search process, allows the algorithm to jump out of the current local optimum faster and searches a larger range for the optimum.Besides, a certain number of single-dimensional learning is applied to the best fireflies, and the learned fireflies guide population evolution.The algorithm proposed in this paper outperforms other upgraded swarm intelligence algorithms and has a good ability to jump out of local optimums, according to experimental results on benchmark test functions.

关 键 词：萤火虫算法 全局最优萤火虫 寻优移动 单维度学习 单增量策略 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]