具有全局收敛性的改进萤火虫优化算法

Research on Improved Glowworm Swarm Optimization with Globally Convergence

对萤火虫优化(Glowworm swarm optimization,GSO)算法全局收敛性及其改进算法性能进行了研究。分析了GSO全局收敛性,针对其收敛效率低的缺陷,提出了一种基于族群划分的改进GSO算法,借鉴混合蛙跳算法思想,将萤火虫群体进行族群划分,局部搜索及全局信息交换的方式改善了算法性能,通过引入萤火虫移动组元概念,改进了萤火虫更新策略,在此基础上,利用混沌优化技术,对萤火虫群体进行初始化,使得算法获得较高质量的初始解群体,并证明了改进算法以概率1收敛于全局最优,最后,采用经典测试函数进行测试,仿真结果表明,改进的萤火虫优化算法在收敛速度及求解精度上有明显改善。

The global convergence of GSO （Glowworm swarm optimization） algorithm and its improved algo- rithm performance are studied. The global convergence analysis of basic GSO is made. In order to improve the GSO convergence efficiency, an improved GSO （IGSO） is presented. Using shuffled frog leaping algorithm （SFLA）, glowworms are divide into different ethnic groups, and local search and global information exchange method im- proves the GSO performance. The introduction of the concept of mobile group element is proposed in order to im- prove glowworm diversity. By using chaos optimization technique, glowworm groups are initialized, and the algo- rithm can obtain high quality initial solutions group. Finally, with the classical test functions, the simulation results show that, the GSO with hybrid behavior has better convergence speed and precision.