一种自适应的微粒群算法

研究发现，种群中个体间交换信息的方式对微粒群算法的性能影响很大。我们定义种群拓扑结构（population topology）为种群内部不同个体之间交流信息的网络。不同的种群拓扑结构有着各自的特点，有些利于加速收敛，有些利于扩展搜索空间。在分析种群拓扑结构变化特点的基础上，提出了一种新的自适应的微粒群算法。和通过调节惯性权重的自适应微粒群算法不同，本算法是通过改变种群拓扑结构来达到自适应优化目的的。

A New Modified Adaptive Particle Swarm Optimization

In our study, the way of communication among individual and its neighbors plays a key role in the performance of particle swarm optimization. Now we give a clear definition for population topology that a communication network among individuals of population. These different population topologies have their own characters, some ones provide a faster search efficiency and the others offer the advantage that subpopulation could search diverse regions of problem spaces. On the basis of these features, we propose a new modified adaptive particle swarm optimization. Different from the traditional APSO, a new method using dynamically changing population topology is presented.