Abstract: Aimed at particle swarm optimization（PSO） algorithm is easy to fall into local optimal problems for optimizing a high-dimensional population, a new cooperative coevolving particle swarm optimization on K-means cluster（KMS-CCPSO） algorithm is put forward. In the proposed algorithm, the subspace of local search range is designed by K-means algorithm, and the new points’ position and velocity in the search space is relied on Cauchy and Gaussian distributions. The experimental results suggest that the proposed algorithm has better optimization performance, its advantage on the large-scale population optimization problem is more apparent.

Key words: cooperative co-evolution, k-means, high-dimensional optimization, particle swarm optimization, local optimal

摘要： 针对粒子群优化（PSO）算法优化高维问题时，易陷入局部最优，提出一种基于K-均值聚类的协同进化粒子群优化（KMS-CCPSO）算法。该算法通过引入K-均值算法扩大种群的局部搜索范围，采用柯西分布和高斯分布相结合的方法更新粒子的位置。实验结果表明，该算法具有较好的优化性能，其优势在处理高维问题上更为明显。

关键词: 协同进化, K-均值, 高维优化, 粒子群优化, 局部最优