CAJ | 학술논문

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.
침대모의퇴화유전산법중불합리체대방식이급고립적비교궤제,제출일충신적기우소생경모의퇴화적유전산법.통과온도적축보강온,가이재진화조기증강충군다양성,이재진화말기가속산법적수렴과정,유효극복료유전산법용역조숙、국부수색능력차적결점.동시산법환사용료최우보류책략체대료륜반도선택산자,종이유효지감소료괄응도상대고적개체재충군중쾌속확산적가능성.연구결과표명:여상견적모의퇴화유전산법상비,신방법능구유효제고유전산법적수렴성능.
@@@@In order to overcome the problems of unreasonable substitutions and isolated comparative mechanism in genetic simulated annealing, a new genetic algorithm is proposed based on niche simulated annealing in this paper. With the temperature reduction, this algorithm provides a better diversity of population at the early evolution stage, and speeds up the convergence process in the later stage. Therefore, it overcomes the premature convergence and the inability of local search in the process of ordinary genetic algorithm. The elitist strategy is used to replace the roulette wheel selection, which effectively reduce the possibility of rapid spreading of the relatively high fitness individuals in the population. The simulation results of standard test functions show that the new method can effectively improve the convergence performance of genetic algorithm comparing to the common genetic simulated annealing.