基于-Stretching-技术的免疫遗传算法地研究.pdfVIP

th Proceedings of the 25 Chinese Control Conference 7-11 August, 2006, Harbin, Heilongjiang 基于“Stretching”技术的免疫遗传算法的研究 洪露，穆志纯 北京科技大学信息工程学院, 北京 100083 E-mail: honglu912@163.com 摘 要: 为了提高人工免疫算法（AIA ）对复杂多模态函数优化问题的全局最优解搜索能力, 本文在免疫原理和遗 传算法机制的基础上引入“Stretching”技术设计了拉伸免疫遗传算法（SIGA）, 该技术能在算法的优化过程中, 不断缩小目标函数极值点的范围,降低算法的搜索难度, 从而提高了算法的全局收敛性能.利用Markov链为数学工 具, 从理论上证明了SIGA以概率1全局收敛. 实验结果表明, 基于该技术的SIGA算法比传统免疫遗传算法（IGA ） 具有更好的收敛性能. 关键词: 人工免疫算法, 函数“Stretching”技术, 多模态函数优化 Study of Immune Genetic Algorithm Based on “Stretching” Technique Hong lu, Mu Zhichun School of Information Engineering, University of Science and Technology Beijing, Beijing 100083 E-mail: honglu912@163.com Abstract: In order to enhance the global optimal search ability of AIA in solving multi-modal function optimization problem, based on mechanism of immune theory and genetic algorithm, a function “stretching” technique is introduced to obtain stretching immune genetic algorithm (SIGA). The algorithm equipped with the technique can narrow the range of extreme value of the objective function, so it reduces searching difficulty and improves the global convergence performance effectively. Using mathematical methods of Markov chains theory, it is proved theoretically that the SIGA is global convergent with probability 1. Simulation results indicate that the SIGA equipped with the “Stretching” technique exhibits better convergent performance compared with traditional immune genetic algorithm. Key Words: Artificial Immune Algorithm, Function “Stretching” Technique, Multi-Modal Function Optimization