动态拓扑结构的多目标粒子群优化算法

Dynamical Topology Multiobjective Particle Swarm Optimization Algorithm

摘要:

介绍了一种动态拓扑结构的多目标粒子群优化算法(dynamical topology multipleobjective particle swarm optimization,DMPSO).给出了一种新的储备集更新策略,定义了支配度和邻域拥挤度及粒子差异度的概念,根据支配度及邻域拥挤度的大小来决定储备集的更新,增强了解的多样性和均匀性.为了防止早熟收敛,结合邻域拥挤度和粒子差异度,给出了一种拟小世界动态拓扑邻域结构来平衡粒子的全局搜索能力和局部搜索能力.最后通过对几个例子的数值实验说明算法的可行性,并通过成功地应用在实际工程问题上说明方法的有效性.

Abstract:

The paper presents a dynamical topology multiobjective particle swarm optimization(DMPSO)algorithm and a definition of the degree of domain and congestion and discrepancy of particles as well as a strategy of renewing archives which depends on the degree of domain and congestion around neighborhood.So the diversity and uniformity of solution are enhanced.In order to overcome the premature convergence,a new imitating small world dynamical topology strategy based on congestion degree and discrepancy degree is applied to balancing the ability of global searching and local searching.In the end,the application successfully to engineering shows that the DMPSO is feasible and effective.