基于代沟信息的自适应遗传算法

Adaptive genetic algorithms based on generational gap information

针对现有自适应遗传算法无法兼顾群体特性 ,难以稳定地收敛到最优解的问题 ,从种群多样性和适应度均值变化的角度 ,分析了进化停滞或退化的原因 .以种群适应度均值和多样性作为概率调整依据 ,提出了一种新的基于种群代沟信息的自适应遗传算法 .利用相邻两代群体间的适应度差异和多样性差异信息 ,设计了遗传概率的自适应调整策略 ,使算法维持较好的多样性 ,有效避免了早熟 .并证明了算法收敛性 .仿真结果表明该算法能够使种群保持良好的可进化性和收敛性 .

Focusing on the problems existed in some adaptive genetic algorithms that cannot take acc ount of population features and are difficult to converge to optimal solution s tead ily, the reasons of stopping evolution or degeneration are analyzed in view of f itness average and population diversity. According to the fitness average and population diversity, a novel adaptive genetic algorithms based on generatio nal gap information is proposed. By using the information of fitness difference and diversity difference between two neighbour generations, a strategy for genet ic probability adjusting is designed. By this adaptive genetic algorithm, good d iversity can be remained and premature can be avoided effectively. Convergence o f the algorithm is proved. Simulation results show that the proposed algorithms can make the population remain good evolutivity and convergence.