An enhanced firefly algorithm for function optimisation problems

Abstract

An enhanced firefly algorithm (EFA) is proposed to improve the performance of the standard firefly algorithm (FA). After analysing the impacts of the attraction and the randomisation parameter in FA’s movement, we introduce three improvement strategies. First, a virtual standard space is constructed to eliminate the negative influence generated by the long distance between two fireflies. Second, the randomisation parameter in FA’s movement is replaced by a stochastic factor, which is integrated into FA’s attractiveness to produce adaptive randomisation. Finally, a fine-grained evaluation strategy is applied to deal with the negative mutual influence among different dimensions. The experiments carried on classic and shifted benchmark functions show that the proposed EFA performs significantly better than standard FA in terms of both convergent speed and solution precision.