基于种群和荧光素自然感应的萤火虫算法  

Multimodal Glowworm Swarm Optimization based on Species and Natural Sensed Luciferin

作　　者：张玉丽[1] 梁波[1] 

机构地区：[1]大连交通大学理学院,辽宁大连116028

出　　处：《大连交通大学学报》2015年第6期111-116,共6页Journal of Dalian Jiaotong University

基　　金：国家自然科学基金资助项目(61303183;11201045)

摘　　要：针对基本萤火虫算法优化多模函数时计算复杂度高和需要预先设定较多参数值的问题,提出了两种修正的萤火虫算法:基于种群的萤火虫算法(S-GSO)和基于荧光素自然感应的萤火虫算法(LNS-GSO).这种改进使学习行为更符合自然界生物的学习规律,更有利于萤火虫发现问题的所有局部最优解.通过6个标准测试函数测试,结果表明结合运用这两种修正的萤火虫算法能够取得良好的收敛性,在寻找多模函数的峰值个数上显示出较强的优势.Aiming at the problem that Glowworm swarm optimization （GSO） has high computational complexi- ty and too much pre-specified parameters for solving the muhimodal functions, two modified GSO： S-GSO （Species based Glowworm Swarm Optimization） which has low computational complexity and LNS-GSO （Lu- ciferin Natural Sensed Glowworm Swarm Optimization） which does not rely on any prespecified parameters and can form niches automatically are proposed. In these ways, the glowworm behavior accords with the biological natural law even more, and multiple optima of a given muhimodal function can be easily found. Simulation ex- periments on six standard multimodal functions are carried out. The results show that a combination with the two modified optimization strategies has nice convergence ability, and the modified GSO algorithms perform very well in capturing multiple optima of muhimodal functions.

关 键 词：多模函数优化 蚁群优化 粒子群优化 萤火虫群优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]