Study On Dynamic Of Complex Networks In Evolutionary Algorithm

Evolutionary algorithms which based on Darwinism use the survival of the fittest to simulate problem solving.After the evolution of several generations,finally obtaining is fit for the problem optimal solution.The evolutionary algorithms started earlier,development mature,and widely applied in various fields of society.But the existing study only focus on the ultimately results or predicted results,it often ignores the relationship between individuals in evolution process and the relationship changes over the evolutionary,and the influence of the changing relationship on the result and the rate of convergence.In this paper,we study these neglected problems.By studying the changing relationship between individual,the structure of complex networks and complex network dynamics are studied.With the continuous development of complex network theory,its application has become more and more popular.The optimization process of algorithm can be described by network dynamics.In the process of iteration,the total number of individuals,in which participate the evolution,remains the same,namely,the total number of nodes in structure remains the same.Network structure is changed in the connection probability.Real network structure directly affects the dynamics of network,dynamics on complex networks may in turn affect the network structure.The complex interaction and effect between network structure and dynamics process is called "co-evolution",it has become a hotspot research in recent years.The dynamical processes of complex network can be used to describe the optimization process of evolutionary computation.The relationship between this two fields will be a hot topic in the future.In this thesis,two completely different(at first glance)areas of research are discussed,complex network structure whether hidden in evolutionary algorithm.First,the evolutionary algorithm and the optimization process of improved algorithm are studied.Second,the interrelationships among individuals in the process of optimization are analyzed.Third,the relationship among individuals in complex network model can be described by dynamic processes,and its complex network structure is discussed.Finally,the influences of the changing relationship to the evolution results and convergence speed are analyzed,and the improving suggestions are presented to improve algorithm.The experimental results show that optimization process of evolutionary algorithm can be described by complex network dynamics.Evolutionary algorithm could be controlled or improved by complex network technology.This research has important theory and practice significances for the further study of complex networks and the improvement,optimization and control of evolutionary algorithm.