A Novel Firefly Algorithm based on Improved Learning Mechanism

Abstract

According to the problem of low solution precision, slow convergence speed and some fireflies failure in the traditional firefly algorithm, an improved firefly optimization algorithm is proposed based on the learning mechanism, By chaotic maps, the Firefly's initial position in which dynamic distance weighting is introduced to strength the search ability of algorithm, and adaptive-step scheme is used to balance the dual requirement of local and global optimization. Gaussian mutation is also adopted in the algorithm to help firefly population jump out of local optimum effectively. Experimental results show that the proposed algorithm has rationality and effectiveness.