Improved And Applied Research Of Firefly Algorithm

The firefly algorithm (FA) is a novel meta-heuristic algorithm that is proposedby Xin-she Yang who comes from University of Cambridge, which is inspired bythe phenomenon of bioluminescent communication and the social behavior such asfeeding, choosing spouse and warning mechanism etc., the algorithm solvesoptimal problem by the fireflies’ attraction mobile. So far, firefly algorithm hasbeen successfully applied to combinatorial optimization, path planning, imageprocessing, and economic dispatch, and show good performance and the potentialfor practical application. However, the firefly algorithm is put forward in recentyears, as a new swarm intelligent optimization algorithm, there are many problemsare worth studying, such as easy to fall into local optimum, slow convergence inthe late iteration, lack of mathematics theory foundation, and its application rangeis relatively narrow, so it is very necessary to do more research and analysis on itsrule and performance, and expand its application field. Based on this, this thesiscarries out the following research:(1) The Firefly Algorithm (FA) has a few disadvantages in the globalsearching including slow convergence speed, high possibility of being trapped inlocal optimum and low solving precision, an improved FA based on simplexmethod is proposed, which combines the characteristics of speedy local search ofsimplex method with the global optimization of firefly algorithm. Theexperimental result shows that, the improved algorithm has improved significantlyin function optimization.(2) By integrating the greedy and mutation strategy into firefly algorithm, agreedy firefly algorithm is proposed to solve0-1knapsack problem, it could makethe firefly jump out of local optimum to a certain extent through adding the greedyand mutation strategy, thus improve the optimization performance of the algorithm.The experimental result shows that, the algorithm of this thesis has strongerconstraint handing ability and higher rate of convergence for solving0-1knapsackproblem. (3) The cooperative search technique is added to firefly algorithm, it couldimprove the ability of fireflies jump out of local optimum through cooperativesearch technique. Through the experimental simulation of some UCI date sets, theresult shows that, the algorithm of this thesis has a better effect on clusteringanalysis.(4) The impact of the key parameters in the algorithm to the optimizeperformance of the algorithm is discussed, and use some standard test functions forthe simulation experiment, the effect of the key parameters in the process ofoptimizing is verified by the experimental result.