摘要
萤火虫算法(FA)是一种基于群智能的优化技术,它在很多优化问题上表现出较好的性能。然而,它求解复杂优化问题时存在一些问题,如收敛速度慢,精度低。针对这些问题,本文提出了一种新的萤火虫算法(取名AFA),该方法使用了三种混合策略,以获得好的优化性能。它首先使用一种自适应的参数方法来动态改变步长参数,然后应用一种改进的搜索策略来消除吸引力,于是,AFA不再包含光吸收系数和初始吸引力这2个参数;再使用反向学习来提高解的精度。仿真结果表明,本文提出的AFA算法优化结果优于MFA及PAFA算法。
Firefly algorithm (FA) is a recently proposed optimisation technique, based on swarm intelligence, which has shown good optimisation performance. However, FA suffers from slow convergence and low accuracy of solutions. To improve this case, this paper presents a new firefly algorithm (AFA) by using three hybrid strategies to obtain a good optimisation performance. First, an adaptive parameter method is used to dynamically changing the step factor. Second, AFA uses a modified search strategy and eliminates the concept of attractiveness. So, HFA does not include two parameters, ab-sorption coefficient and initial attractiveness. Third, a concept of opposition-based learning is used for improving the accuracy of the global best solution. Experiments on some benchmark problems show that AFA is superior to mimetic FA (MFA) and probabilistic attraction-based FA (PAFA).
出处
《计算机科学与应用》
2020年第9期1639-1645,共7页
Computer Science and Application
关键词
萤火虫算法
自适应搜索策略
反向学习
优化
Firefly Algorithm
Adaptive Search Strategies
Opposition-Based Learning
Optimisation
