【摘要】 受古代"部落通婚"现象启发,提出了一种新的多种群遗传算法的种群间迁移策略,并以此构造了一种采用"部落通婚"策略的遗传算法。为了验证该算法的有效性,考察了分类BP网络的训练问题和典型测试函数的优化问题。模拟结果表明,这种改进的遗传算法能有效地避免"早熟"现象,其收敛精度和收敛速度比标准遗传算法和通常的多种群遗传算法有一定程度的提高。更多还原

【Abstract】 Inspired by the tribe-intermarry phenomenon in ancient society, a novel improved genetic algorithm using immigration strategy among populations of the multiple-population is presented. Based on the strategy, a TISGA(Genetic Algorithm with Tribe-Intermarry-Strategy) is proposed. In order to validate the algorithm, the training of BP neural network for classification and the optimization of typical benchmark functions are examined. Simulation results show that the improved algorithm could avoid premature efficiently.It is superior to the standard GA(Genetic Algorithm) and the multiple-population GA in both accuracy and speed of convergence.更多还原