论文部分内容阅读
针对模拟退火遗传算法中不合理替代方式以及孤立的比较机制,提出一种新的基于小生境模拟退火的遗传算法.通过温度的逐步降温,可以在进化早期增强种群多样性,而在进化末期加速算法的收敛过程,有效克服了遗传算法容易早熟、局部搜索能力差的缺点.同时算法还使用了最优保留策略替代了轮盘赌选择算子,从而有效地减少了适应度相对高的个体在种群中快速扩散的可能性.研究结果表明:与常见的模拟退火遗传算法相比,新方法能够有效提高遗传算法的收敛性能.
Aiming at the unreasonable substitution methods and isolated comparison mechanism in simulated annealing genetic algorithm, a new genetic algorithm based on niche simulated annealing is proposed, in which the population diversity can be enhanced early in evolution and accelerated at the end of evolution The convergence of the algorithm effectively overcomes the shortcomings of genetic algorithm, such as easy premature, poor local search ability, while the algorithm also uses the optimal retention strategy to replace the roulette wheel selection operator, so as to effectively reduce the individuals with relatively high fitness The possibility of rapid spread in the population.The results show that compared with the common simulated annealing genetic algorithm, the new method can effectively improve the convergence performance of the genetic algorithm.