一种结合信赖域算法的混合MIMIC算法  

A Hybrid MIMIC Algorithm Integrated with Trust-region Algorithm

作　　者：夏桂梅[1] 张文林[1] XIA Guimei ZHANG Wenlin(School of Applied Science, Taiyuan University of Science and Technology, Taiyuan, China 03002)

机构地区：[1]太原科技大学应用科学学院,山西太原030024

出　　处：《温州大学学报（自然科学版）》2017年第2期1-7,共7页Journal of Wenzhou University(Natural Science Edition)

基　　金：山西省自然基金(2014011006-2);太原科技大学研究生教改项目(20133001)

摘　　要：针对分布估计算法在求解问题的过程中局部搜索能力较弱的缺点,引入了信赖域算法,提出了结合信赖域算法的分布估计算法.由于信赖域算法是一种很好的局部快速寻优方法,因此在分布估计算法的基础上,再对每一个粒子分别实施信赖域算法,能够加强算法的局部搜索能力.新算法不仅保持了种群的多样性,而且具备更全面的学习能力,提高了算法的寻优能力,避免早熟收敛的发生.数值试验结果表明:该算法能收敛到满足约束条件的最优解,并且具有很强的搜索能力,为解决非线性约束优化问题提供了一种新的有效途径.The Trust-region algorithm is introduced in this paper in allusion to the defect with the estimation of distribution algorithms which is weak in local search capability. Meanwhile, the distributed estimation algorithm integrated with the trust-region algorithm is proposed. Due to the trust-region algorithm is a partial optimal seeking method, the local search capability of the trust-region algorithm is strengthened with each particle implemented respectively. The new algorithm not only maintains the diversity of population, but also possesses a more comprehensive learning ability, which improves the searching capability of the algorithm, and avoids the occurrence of premature convergence. Numerical experiments show that the proposed algorithm is in position to converge the optimal solution which meets every constraint condition. Therefore, it is a very strong searching ability and provides a brand-new effective way to solve nonlinear constrained optimization problems.

关 键 词：分布估计算法 信赖域算法 MIMIC算法 

分 类 号：O221[理学—运筹学与控制论]