摘要
分布估计算法是基于遗传算法的基础上发展起来的一种新型的优化算法,它采用的是概率图的模型来表示基因中变量之间的关系,从而构建优良解集的概率分布模型进行采样来实现迭代进化。但是分布估计算法在问题求解过程中容易陷入局部最优,针对此缺点,引入微粒群算法,提出了一种结合微粒群算法的分布估计算法。这种算法将分布估计算法与微粒群算法的思想紧密结合起来,不仅保持了种群的多样性,而且具有更全面的学习能力,提高了算法的寻优能力以及避免早熟收敛的能力。通过对测试函数的仿真试验,表明该算法具有良好的性能。
The estimation of distribution algorithm is developed on the basis of genetic algorithm; it is a kind of new optimization algorithm which uses probability graph model to express the relationship among variables to construct the gene so as to build excellent solution set of probability distribution model for sampling to realize iterative evolution. But the estimation of distribution algorithm easily falls into local optimum in the problem solving process. Aiming at this disadvantage,the particle swarm algorithm is proposed to estimate algorithm combined with the distribution of particle swarm optimization algorithm. The new algorithm combines the estimation of distribution algorithm with particle swarm algorithm,keeps the diversity of population,has more comprehensive learning ability and improves searching ability to avoid premature convergence. Through simulation test of test function,the algorithm is proved to have good performance.
出处
《太原科技大学学报》
2015年第5期406-410,共5页
Journal of Taiyuan University of Science and Technology
基金
山西省自然基金(2014011006-2)
关键词
分布估计算法
微粒群算法
MIMIC算法
estimation of distribution algorithms
hybrid particle algorithm
MIMIC algorithm