聚类分析算法改进及其在地铁隧道病害分析中的应用研究

在研究经典聚类分析算法的基础上,对k-means算法进行改进,得到MK-means改进算法。随机选取初始点,运算后会产生不同的聚类结果。把基于图论中最小支撑树思想应用到算法改进中,通过最小支撑树求得最小圈,产生k个初始聚类中心,由达到一定初始聚类阈值的点开始聚类,聚类效果更优,有效克服随机选取初始点的缺点。把改进的算法应用到地铁隧道病害分析评价中,准确划分隧道病害等级,有针对性地提出防治措施,具有一定的实际应用价值。

Improvement of Clustering Analysis Algorithm and the Application in Tunnel Disease

Classical algorithm based on cluster analysis is going to improve the k-means algorithm,so as to get the MK-means Algorithm.In a random selection of initial point iteration,each operation will produce different clustering results.Based on the minimum spanning tree of graph theory and by a calculation,the minimum cycle will be obtained,and the beginning ofk-start cluster center will be produced.When up to a certain point,the initial clustering threshold will start clustering,which will have some practical value for the tunnel disease prevention.