基于粒子群优化的带障碍约束空间聚类分析

聚类分析是空间数据挖掘的主要方法之一.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.在分析K中心聚类方法易陷入局部极小值和对初始值敏感的基础上,提出了一种新的聚类方法--基于粒子群优化的带障碍约束空间聚类方法.实验结果表明,该聚类方法不仅使得聚类结果更具实际意义,而且在全局寻优能力方面明显优于K中心聚类方法,且有较快的收敛速度.

PSO spatial clustering with obstacles constraints

Clustering spatial data is one of the main methods in the spatial data mining (SDM). Classical clustering algorithms have ignored the fact that many constraints exit in the real world and could affect the correctness of clustering result. After analyzing the disadvantages of the classical K-medoids clustering algorithm, a novel K-medoids clustering algorithm based on particle swarm optimization algorithm with obstacles constraints is proposed. The classical K-medoids algorithm exits the local optima and is sensitive to initialization. The experimental results show that this novel clustering method not only has greater searching capability, but also has fast convergent rate.