基于改进MOEA/D算法的WSN覆盖优化方法  

Improving coverage of wireless sensor network using enhanced MOEA/D algorithm

作　　者：神显豪[1] 李军[1] 张祁[1] 

机构地区：[1]桂林理工大学信息科学与工程学院,广西桂林541004

出　　处：《计算机应用研究》2016年第4期1203-1206,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(E050603);广西高等学校科研资助项目(YB2014157);广西自然科学基金资助项目(2015GXNSFBA139254)

摘　　要：为了优化无线传感器网络(WSN)的覆盖方法,针对MOEA/D中缺少对本代优质个体的保存和最优解集中个体极少的两个问题,提出了粒子群优化的基于分解的多目标进化算法(MOEA/D-PSO)。通过保留种群本代优质个体,改进本地优化解集在进化过程中的搜索方向和搜索进度,弥补了MOEA/D的不足。仿真实验证明,相对于MOEA/D和非支配排序遗传算法(NSGA-Ⅱ),MOEA/D-PSO所得非支配解更接近Pareto最优曲面,解集分布的均匀性和多样性表现更佳,WSN的覆盖范围更广,能量消耗更少。In order to optimize the coverage of wireless sensor network（ WSN）,for the MOEA / D had two deficiencies,such as the lack of preserving generation of high quality individuals and individuals rarely in optimal solutions,this paper proposed MOEA / D-PSO（ multi-objective evolutionary algorithm based on decomposition particle swarm optimization）. By preserving the generation of high-quality individuals population,and improving local optimization solution set in evolutionary search direction and the search for progress,it made up for the shortcomings of the original MOEA / D. Simulation experimental results show that compared with MOEA / D and NSGA-Ⅱ（ non-dominated sorting genetic algorithm-Ⅱ）,the non-dominated solutions of MOEA / D-PSO is closer to Pareto optimal surface,uniformity and diversity of solution set distribution perform better. Coverage of WSN is more widely,and consumes less energy.

关 键 词：基于分解的多目标进化算法 粒子群优化 帕累托最优曲面 覆盖范围和能量消耗 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]