摘要
针对传统克隆选择算法在变异时存在的盲目性和随机性而导致的退化现象以及容易陷入局部最优等问题,本文以生物免疫机制中抗体是由稳定区和可变区构成为理论支撑,利用粗糙集中的核值概念,提出了一种基于粗糙集核值的克隆选择算法.该算法用核值构造稳定区,使每一代优秀抗体的核信息在成熟的过程中保持不变,只对可变区进行变异操作,使得变异具有稳定性和向最优解靠近的方向性.实验部分采用经典的测试函数对两种算法的性能进行测试对比.实验结果表明该算法在收敛速度、抗体多样性以及避免早熟等方面均比传统克隆选择算法具有更好的效果,且算法在迭代后期还具有较强的局部搜索能力.
The traditional clonal selection algorithm easily fell into the local optimum and was led to the degeneracy phenomenon for the blindness and randomness in the process of the mutation. Noticing that the antibody structure consists of the stable region and the variable region in the biological immune mechanisms, the core value of the rough set theory was employed to imitate the stable region of the antibody structure. Based on the core value, an improved clonal selection algorithm was proposed, where it kept the core informarion of the superior antibodies unchanged in each generation of the maturation process, and only the highly variable region was operated. The proposed algorithm makes the mutation process more stable and with the way to the optimal antibody. The experiment part adopts the classical testing functions tests the performance of the two algorithms. The results show that this proposed algorithm is better than the traditional clonal selection algorithm in the convergence speed, the antibodies diversity, avoiding the premature and the strong local searching ability in the late progress of the iteration.
出处
《小型微型计算机系统》
CSCD
北大核心
2016年第5期992-996,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61450011)资助
山西省自然科学基金项目(2014011018-2)资助
山西省回国留学人员科研项目(2013-033)资助
山西省留学回国人员科技活动择优项目资助
关键词
克隆选择
人工免疫
粗糙集
核值
数值优化
clonal selection
artificial immune
rough set
core value
numerical optimization
