【摘要】 函数优化是遗传算法应用的一个方面,标准遗传算法通常采用的是轮盘赌选择、单点交叉和变异等基本操作算子,其缺点是全局收敛性差,易造成“不成熟”收敛现象。研究表明,GA的收敛性主要是由选择算子实现的,轮盘赌选择易产生较大的随机误差,基于期望值和轮盘赌的混合选择策略则能够改善此误差。仿真结果表明,混合选择能够有效地提高GA对全局最优解的搜索能力,较好地改善“早熟”现象的产生。更多还原

【Abstract】 Function optimization is one of the applications of genetic algorithm(GA).Commonly,three operators including roulette selection,single crossover and mutation are adopted in simple genetic algorithm(SGA).The shortcoming of SGA is bad global convergence,with possible premature convergence. The study shows that the convergence of GA mostly lies in the selection operator.But the roulette selection has more random error.In this paper,hybrid selection strategy based on the expected value and roulette wheel selection is proposed to improve the demerit.The simulation results show such a strategy can effectively improve the ability of searching the global optimum solution and avoid premature convergence.更多还原