A Randomly Guided Firefly Algorithm Based on Elitist Strategy and its Applications

Abstract

Firefly algorithm (FA) is one of the swarm intelligence algorithms, which is proposed by Yang in 2008. The standard FA has some disadvantages, such as high computational time complexity, slow convergence speed and so on. The main reason is that FA employs a full attracted model, which makes the oscillation of each firefly during its movement. To overcome these disadvantages, based on elitist strategy, a randomly guided firefly algorithm (ERaFA) is proposed. In this algorithm, for improving the convergence speed, an elitist attraction model is developed based on random selection from elite fireflies, which can lead the firefly to a right direction. To deal with the possible failure of the elite guidance, opposite learning strategy is adopted. Meanwhile, to strengthen the local search ability of our algorithm, and help our algorithm jump out a local optimum position, a new mechanism is proposed, which is similar to the crossover operator in GA. The performance of ERaFA is evaluated by some well-known test functions and applied to solve three constrained engineering problems. The results show that ERaFA is superior to FA and some other state-of-the-art algorithms in terms of the convergence speed and robustness.