Study on the Random Factor of Firefly Algorithm.

The firefly algorithm (FA) is a swarm intelligence algorithm that mimics the swarm behaviour of the firefly in nature. The idea is simple, and FA is easy to realize. To improve its performance, a new method to control the random factor in FA is proposed in this paper, based on the design idea and mathematical model of FA and a simple experiment. Under the new method, the value of the random factor decreases according to a geometric progression sequence. Twenty common ratios of geometric progression sequences are used to optimize nine standard benchmark functions. The experimental results are analysed by the ANOVA and step-up methods. The analysis shows that the performance of FA improves under the new method to control the random factor.