Abstract: By introducing K-means for the parallel particle swarm optimization based on an island model，the populations are divided into several sub-populations.It can not only make the location of the particles in the same sub-population be in the relative concentrative，and be relatively easy to learn，but also improve the search efficiency，so that the limited time will be spent on the most effective search.According to the characteristics of them，the parallel algorithms are improved.When certain conditions are met，it carries on communications，so that ineffective communications can be avoided to reduce the time spent for communications.The simulation results confirm that the algorithms have a high convergence speed and convergence accuracy.

Key words: K-means, parallel, particle swarm optimization, optimization

摘要： 通过给基于孤岛模型的并行粒子群算法引入K-means来进行子种群的划分。这不仅可以使一个子种群中的粒子位置相对集中，学习相对容易，而且可以提高搜索效率，使有限的时间用在最有效的搜索上。针对并行算法的特点，对其进行改进，在满足一定条件时才进行通信，这样可以避免无效通信，减少通信所花的时间。仿真结果证实，该算法具有较高的收敛速度和收敛精度。

关键词: K-means, 并行, 粒子群, 优化