改进的免疫优化算法对动态约束多目标问题的应用  

APPLYING IMPROVED IMMUNE OPTIMISATION IN DYNAMIC CONSTRAINT MULTIOBJECTIVE PROBLEM

作　　者：武慧虹[1] 钱淑渠[1] 高忠生[1] 

机构地区：[1]安顺学院数学与计算机科学系,贵州安顺561000

出　　处：《计算机应用与软件》2014年第4期293-296,319,共5页Computer Applications and Software

基　　金：贵州省科学技术基金项目(黔科合J字20122002);贵州省教育厅自然科学基金项目(20090074)

摘　　要：基于进化理论的动态多目标优化算法极易陷入局部最优,跟踪动态Pareto有效面的速度及效果较差。基于免疫系统机理提出一种改进的免疫优化算法(DMIOA)用于动态约束多目标问题求解。算法通过抗体浓度及其支配度设计抗体与抗原亲和力,随机约束选择算子提高算法约束处理能力,环境识别算子自适应判断环境变化,根据识别结果以不同的方式产生新环境的初始抗体群。数值实验中,将DMIOA应用于两种动态标准测试问题及飞机减速器参数动态设计问题的求解,结果表明:DMIOA能快速跟踪动态Pareto有效面,且在各环境所获面分布均匀,具有较好的实际问题求解能力。Dynamic multiobjective optimisation algorithm based on evolution theory is extremely easy to fall into local optimum,and is poor in both speed and effect of tracking the dynamic Pareto effective front. We propose an improved dynamic immune optimised algorithm（ DMIOA） based on immune system mechanism for solving the dynamic constraint multiobjective problem. Through antibody＇s density and its dominated degree,the algorithm designs the affinity of antibody and antigen,the random constraint selection operator to improve the ability of constraint and processing,and the environments recognition operator to adaptively examine the environmental changes,and generates initial antibody population in new environment in different ways according to the recognition result. In numerical experiments,DMIOA is applied to the solutions of two kinds of dynamic standard test problems and the dynamic design problem of airplane speed reducer parameters,the results indicate that the DMIOA can rapidly track the dynamic Pareto effective front,and is evenly distributed on each obtained environmental surface,it has desirable capability on solving practical problems.

关 键 词：动态环境 多目标优化 免疫算法 动态Pareto有效面 

分 类 号：TP306.21[自动化与计算机技术—计算机系统结构]