应用小种群远缘杂交提高遗传算法的效率

Improving the efficiency of genetic algorithm with distant hybridization among micro populations

为提高遗传算法的效率,将作物育种学中远缘杂交策略应用于多种群遗传算法,采用规模较小的多个种群同时进行进化,选择和变异操作在各种群内部独立完成,杂交操作在种群间完成。小种群的采用可以大大提高进化求解的速度,种群间的远缘杂交能够克服由于种群规模小、种群多样性降低导致早熟收敛的弊端,保证算法以较快的速度收敛到全局最优解。研究结果表明:该算法具有高效性。

For the purpose of improving efficiency of genetic algorithm （GA）, a distant hybridization originated from crop breeding is proposed to apply on multi-population genetic algorithm. Each population with a small size evolves simultaneously. During their evolution, the selection and mutation are completed inside other population and the crossover is completed through the cooperation of multiple populations. The scheme of micro population can greatly reduce the computing time. Distant hybridization crossover can overcome the drawbacks of premature due to the small population size and low individual diversity. Hence the algorithm can converge quickly to obtain a global optimal solution. A case study shows that the proposed genetic algorithm is highly efficient.