Abstract: K-medoids algorithm as a kind of clustering algorithm, not easily affected by extreme data, the influence of broad adaptability, but K-medoids clustering algorithm accuracy is not stable, average accuracy, low in the real, clustering analysis effect is poorer. ACO is a bionic optimization algorithm, which has strong robustness, can be unified easily with other method, has high efficiency. K-medoids clustering algorithm based on ACO algorithm merit reference, this paper proposes a new clustering algorithm. It raises the clustering algorithm, and the stability of the accuracy is high. Finally, simulation experiments show the feasibility and advantage of this algorithm.

Key words: Ant Colony Optimization（ACO） algorithm, cluster analysis, K-medoids algorithm

摘要： K-medoids算法作为聚类算法的一种，不易受极端数据的影响，适应性广泛，但是K-medoids聚类算法的精确度不稳定，平均准确率较低，用于实际的聚类分析时效果较差。ACO是一种仿生优化算法，其具有很强的健壮性，容易与其他方法相结合，求解效率高等特点。在K-medoids聚类算法的基础上，借鉴ACO算法的优点，提出了一种新的聚类算法，它提高了聚类的准确率，算法的稳定性也比较高。通过仿真实验，验证了算法的可行性和先进性。

关键词: 蚁群优化算法（ACO）, 聚类分析, K-medoids算法