中文摘要：

针对K-means算法对随机选择的初始聚类中心敏感且聚类结果不稳定、准确率不高的问题,提出一种基于邻域数据距离加权的聚类中心鲁棒优化算法。通过建立数据密度约束将聚类中心优化在数据密集区域,有效克服K-means算法聚类结果稳定性差等问题。通过对仿真数据和标准数据集的实验,验证了采用该算法收敛的聚类中心非常接近标准数据集的实际中心,具有较优的聚类准确性、鲁棒性和收敛速度。

英文摘要：

For the issue that K-means algorithm is sensitive to its randomly chosen initial clustering centers, and the clustering results are unstable and the accuracy rate is not very high, a robust clustering center optimal algorithm based on weighted neigh- borhood distance was proposed. The algorithm optimally focused the clustering centers on data dense areas, overcoming problems of K-means algorithm including poor stability, etc. Experimental results from simulation data as well as standard data sets show that the convergent clustering centers closely approximate to the actual centers of various types of data sets with high clus- tering accuracy, strong robustness and high convergence rate.