Directed firefly algorithm for multimodal problems

Abstract

Firefly Algorithm (FFA) is very recently introduced one of the Swarm Intelligence (SI) algorithms, developed by the inspiration from flashing light characteristics of fireflies. This algorithm assumes each firefly as potential solutions and brightness associated with them depends on their performance over optimizing problem. The Swarm of firefly get attracted towards the goal by following the brighter firefly and if no such firefly is found, they will move randomly. The basic FFA algorithm follows classical update strategy of attraction and movement of the swarm. This strategy may not produce good quality of results and seems to converge prematurely. The basic movement strategy of firefly may be modified and searching quality of FFA can be improved. This paper proposes novel fireflys movement update strategy. The basic FFA algorithm move the firefly randomly if brighter firefly is not found, this flaw is attended in this paper and thus proposed Directed Firefly Algorithm (DFA). The DFA directs the randomly moving firefly to search around the brightest firefly in the current iteration. To preserve the diversity and avoid premature convergence, the directing strategy is followed at certain refresh rate. The complete performance comparisons of the proposed algorithm are validated against various SI algorithms over standard test cases. The test suite comprises of complex, multimodal and scalable optimization problems, whose dimensions are varied form 10, 30 and 50. The proposed strategy has remarkably improved the quality of solution and convergence rate of proposed algorithm.