摘要
为了克服原始萤火虫算法(Glowworm swarm optimization,GSO)对于高维、多峰函数寻优精度不高和后期收敛速度较慢的问题.为此,有针对性地提出了一种改进的动态步长萤火虫优化算法,在整个迭代期间,萤火虫的算法步长呈非线性递减.在寻优初期保持着一个相对较大的步长进行搜索,增强其全局寻优能力,在算法后期保持一个较小的移动步长,增强其局部搜索能力.此外将原始萤火虫算法在Nit集合为0时不移动,改成试探性向外随机移动策略.采用该算法的改进思想,能在一定的程度上避免算法因为过早的成熟而陷入局部最优,并且改进的算法比原始萤火虫算法有着更好的收敛精度.通过与原始GSO和改进算法ASGSO做对比,Matlab实验仿真表明,此改进算法在一定程度上提高了收敛速度和寻优精度.
In order to overcome the basic artifical firefly algorithm（GSO） in solving problems of low precision for themulti peak function, high dimension and slow convergence. Thus, this paper comes up with a Dynamic stepoptimization algorithm. During the iteration , the step decrease nonlinearly. At the beginning of the optimizationmaintains a large step to search optimal value , enhancing global optimization ability. At the ending of it ,optimization maintains a small step to improve ablity of local searching. In addition, original firely algorithm does notmove anywhere, when Nit is null. this paper makes it move somewhere randomly. With it, This algorithm can avoidbeing mature and falling into a local value. As the same time, improved GSO can achieve a higher accuracy duringIterative process. , comparing with GSO and reference literature alogorithm ASGSO. According to the simulationexperiment, it shows that to some extent, the improved algorithm on covergence speend and precision are enhanced.
出处
《微电子学与计算机》
CSCD
北大核心
2018年第2期61-66,共6页
Microelectronics & Computer
