Abstract: Various clustering methods based on the behavior of real ants are proposed.In this paper，a new algorithm is developed which the behavior of the artificial ants is governed by fuzzy set.The average distance is defined between objects，and the average distance is taken as the similarity of the object domain.Similarity between objects is mapped a domain of fuzzy sets by membership function.By the given confidence level，fuzzy sets will be separated into universal set.The universal set will decide that ants pick up or put down the object.To mine the actual measurement data for the data source，the basic ant colony clustering algorithm and the fuzzy ant based spatial clustering algorithm are used separately.Experimental results prove that the improved algorithm enhances the clustering effect.

Key words: fuzzy set, ant colony, spatial clustering

摘要： 定义了对象间的平均距离，并将平均距离作为对象相似性的论域。通过隶属函数将对象间的相似性映射为论域上的一个模糊子集。由给定的置信水平λ，将模糊集分离为普通集，对蚂蚁是否拾起还是放下对象作出决策，实现对空间数据的聚类。并以矿山实际测量数据为空间数据源，采用基本的蚁群聚类算法和模糊蚁群空间聚类算法分别对其进行聚类。通过对这两种算法的实验结果进行分析比较，证明改进后的算法提高了聚类效果。

关键词: 模糊集, 蚁群优化, 空间聚类