求解多峰函数优化问题的并行免疫遗传算法  

A Parallel Immune Genetic Algorithm for Multimodal Function Optimization Problem

作　　者：程林辉[1,2] 钟珞[2] 

机构地区：[1]中南民族大学计算机科学学院,湖北武汉430074 [2]武汉理工大学计算机科学与技术学院,湖北武汉430074

出　　处：《微电子学与计算机》2015年第5期117-121,共5页Microelectronics & Computer

基　　金：国家自然科学基金面上项目(61379059);湖北省自然科学基金面上项目(2014CFB912)

摘　　要：针对基本遗传算法在求解多峰函数时很难找到全部最优解的不足,提出了一种并行免疫遗传算法求解该问题.算法引入郭涛算法的多父体杂交思想,并借鉴小生境机制将算法的优化过程进行分解.算法进化初期借鉴免疫算法的抗体浓度抑制思想,通过变异算子降低大种群内较优相似个体的浓度,以增加种群多样性,扩大搜索空间,确定各峰的区域.算法后期将种群划分为若干子种群,并通过免疫记忆库,记录各子种群当前的最优解,同时对各子种群进行指导性搜索,以快速收敛到各峰,免疫记忆保留各代的精英个体保证了算法的收敛性.实验结果表明,所提出的算法在求解多峰函数优化问题中取得了满意的结果.For the simple genetic algorithm is difficult to find all the optimal solutions in solving multimodal function optimization problem,aparallel immune genetic algorithm was proposed in this paper.Multi-parent crossover operator of GuoTao algorithm and niche technique was introduced to improve the algorithm.The optimization procedure of the algorithm was divided into two stages.In the early stage of the algorithm,antibody concentration depression of immune algorithm was referenced to increase the population diversity.The concentration of superior similar individual in a large population was reduced by mutation operator to enlarge the search space,thereby determining every modal region.The population was divided into a number of sub-populations in the late algorithm.And immune memory bank was introduced into the sub-populations to record the current optimization of each subpopulation and instructionally help sub-population to converge to each modal quickly.Algorithm convergence was ensured by immune memory keeping the elite individual of generations.The experimental results show that the proposed algorithm takes good performance on the multi-modal function optimization problem.

关 键 词：多峰函数优化 免疫遗传算法 郭涛算法 小生境 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]