Research On Chaotic Firefly Algorithm Based On Roulette Wheel Selection Strategy

With the information technology improving,it is difficult for traditional methods to obtain some desired solutions for complex optimization problems.Swarm intelligence optimization is an emerging meta-heuristic method that can solve optimization problems better in recent years.Some researchers are inspired by the biological evolution,foraging or communication in nature to design the optimization algorithm.The firefly algorithm(FA)is a novel swarm intelligence optimization algorithm that simulates the behavior of fireflies that rely on flashing light for mating or warning off potential predators.It has the advantages of easy to understand,less tunable parameters,higher solving accuracy and convergence rate.It is particularly advantageous for solving continuous and discrete optimization problems.However,there are still some defects in the algorithm,such as high dependence on the distribution of initial population,easy to fall into local optimum,slow convergence rate in the early iteration,always oscillating around the extreme point during the later stage of iteration when addressing complex optimization problems and so on.In this dissertation,the bionic principle and flow of the original FA is investigated firstly and then an improved chaotic firefly algorithm based on roulette strategy(CRSFA)is proposed.The CRSFA improves mainly the original FA from the aspects of initial population,position update formula and optimization process.Firstly,to address the problem of uneven initial population generated by random methods,the improved good-point set method is used to initialize the population,so that the population is evenly distributed in the search space,and at the same time,the convergence speed of the algorithm is improved.Secondly,to address the problem of repeated oscillations around the extreme point in the later iteration,adaptive inertia weight is introduced into the original position update formula to effectively restraining the occurrence of the oscillations problem,which improves the convergence rate and the solution accuracy of the algorithm,and better balances the global search ability and the local search ability.To address the problem of local optimal in the process of optimization,in the framework of the original roulette wheel selection strategy,first of all,according to its brightness from high to low,the former individuals are selected as the elite group.Secondly,the repeated individuals selected by the roulette wheel selection strategy are deleted,and the same number is regenerated by the chaotic sequence with Gauss disturbance to form an improved group.Finally,the elite groups and the improved groups are combined to form a firefly population with a better fitness value and then a position update is performed so as to increase the population diversity,avoid falling into the local optimum and improve global search capabilities.Using eight single-peak functions and seven multi-peak functions,the particle swarm optimization algorithm,the firefly algorithm,a modified FA based on light intensity difference proposed by Yan Xiaohui et al and the CRSFA in this paper are simulated.Experimental studies suggest that CRSFA can effectively improve the shortage of repeated oscillation around the extreme point in the later iteration,avoid falling into the local optimum,improve the accuracy and convergence rate of the algorithm,and balance the global search ability and local search ability.As a result,the proposed CRSFA is shown to be able to outperform the compared algorithms reliably.