Study On The Firefly Algorithm And Application

Firefly algorithm (FA) is proposed by Cambridge scholar Yang Xin-She, it is a stochastic optimization algorithm based on biological swarm intelligence, and it is a new swarm intelligent optimization algorithm following classical intelligent algorithms(such as particle swarm optimization algorithm, genetic algorithm and simulated annealing algorithm), meantime, it is a new branch in the field of evolutionary computation. The algorithm solves optimal problem through the simulation of the social behavior of fireflies, which can realize mobile cooperative behavior by the fireflies’attractiveness during the feeding, choosing spouse,etc. Compared with deterministic optimization method (such as the least square method, gradient algorithm and climbing hill algorithm), the realization of FA is relatively simple, and it has no strict conditions of continuous or differentiable conditions, does not need the prior knowledge, meanwhile, the computational efficiency is higher.However, the firefly algorithm as a relatively new and rapid development of the stochastic optimization algorithm, relative to those algorithms that have strict mathematical background, such as climbing hill algorithm of the deterministic algorithm, there are some key problems that need to be worked out on the systematic and application promotion, the development corresponding hardware products, for instance:algorithm’s premature convergence, easy to fall into local optimum; because of the not appropriate control parameter selection, the algorithm can’t convergence; and how to better solve problem that contains constraint conditions, or problem that contains multiple target optimization, and so on. Considering these disadvantages, the convergence of the algorithm is analyzed in this paper, and on the basis of the key parameters affecting the algorithm to optimize the performance of the algorithm in the analysis of mathematical formula, several experiments are performed, and the optimization calculation function is verified by the experimental result; the inertia weight and chaos theory are introduced in this paper, two kinds of improved algorithms are proposed based on the inertia weight and chaos theory, and they are applied to optimization performance test of the standard test functions, the simulation results show that the improved algorithm not only can ensure faster convergence speed, but also can balance the ability of global searching and local searching, the optimization speed and precision are improved; the algorithm is applied to PID parameters turning and the friction parameters identification problem of the servo system friction based on LuGre friction model, the optimization effect is verified by the specific application problem, the results show that the optimization effect is very ideal, both search speed and accuracy are better than genetic algorithm.