基于改进引力搜索的混合K-调和均值聚类算法研究

为了解决聚类算法容易陷入局部最优的问题，以及增强聚类算法的全局搜索能力，基于KHM算法以及改进的引力搜索算法，本文提出一种混合K-调和均值聚类算法（G-KHM）。G-KHM算法具有KHM算法收敛速度快的优点，但同时针对KHM算法容易陷入局部最优解的问题，在初始化后数据开始搜索聚类中心时采用了一种基于对象多样性及收敛性增强的引力搜索算法，该方法改进了引力搜索算法容易失去种群多样性的缺点，并同时具有引力搜索算法较强的全局搜索能力，可以使算法收敛到全局最优解。仿真结果表明，G-KHM算法能有效地避免陷入局部极值，具有较强的全局搜索能力以及稳定性，并且相比KHM算法、K-mean聚类算法、C均值聚类算法以及粒子群算法，在分类精度和运行时间上表现出了更好地效果。

A mixed K-harmonic means clustering algorithm with improved gravitational search

In order to solve the clustering algorithm is easy to fall into local optimum, and enhanced global search ability clustering algorithm based on gravitational KHM algorithm and improved search algorithm, a mixed K-harmonic means clustering algorithm is proposed. G-KHM KHM algorithm has the advantage of fast convergence, but for the KHM algorithm easy to fall into local optimal solution, after initializing the data center at the start of the search cluster search algorithm uses an object-gravity diversity and convergence enhancement is based. This method improves the shortcomings of gravitational search algorithm is easy to lose the diversity of the population, and also has a strong gravitational search algorithm global search capability, allows the algorithm to converge to the global optimal solution. Simulation results show that, G-KHM algorithm can effectively avoid getting into local extremes, with a strong global search capability and stability, and compared to the KHM algorithm, K-mean clustering algorithm, C-means clustering algorithm and particle swarm optimization in the classification accuracy and run time showed a better effect.