摘要： 空间聚类分析是空间数据挖掘中的一个重要研究课题。传统聚类算法忽略了真实世界中许多约束条件的存在，而约束条件的存在会影响聚类结果的合理性。讨论了带障碍约束的空间聚类问题，研究了一种基于遗传和划分相结合的带障碍约束空间数据聚类分析方法，设计了一个带障碍约束的遗传K中心空间聚类分析算法。对比实验表明，该方法兼顾了局部收敛和全局收敛性能，考虑到了现实障碍物对聚类结果的影响，使得聚类结果更具有实际意义，其结果优于传统K中心聚类及单纯的遗传聚类，不足之处是其计算速度相对较慢。

关键词: 空间数据挖掘, 空间聚类, 遗传算法, K中心算法, 障碍约束

Abstract: Spatial clustering is an important research topic in the spatial data mining(SDM). Classic clustering algorithms ignores the fact that many constraints exit in the real world and can affect the correctness of clustering result. This paper discusses the problem of spatial clustering with obstacles constraints and proposes a novel genetic K-medoids spatial clustering with obstacles constraints based on the genetic algorithm and the K-medoids method. The comparison proves that the method can not only give attention to local constringency and the whole constringency, but also consider the obstacles that exit in the real world and make the clustering result more practice. The experimental results show that it is better than simple genetic algorithm and tradition K-medoids method. The drawback of this method is comparably slower speed in clustering.

Key words: Spatial data mining, Spatial clustering, Genetic algorithm, K-medoids algorithm, Obstacles constraints