摘要
聚类分析是研究类型划分与类间关系的一种数理统计方法。本文将K均值动态聚类分析方法应用于土样识别中。在介绍K均值聚类方法基本思想和计算步骤的基础上,选取某工程勘察的49个土样、每个土样取七个工程性质参数进行K均值动态聚类分析,通过比较样本聚类结果,分析了样本容量对聚类结果的影响,选取模糊C均值的聚类结果,验证了K均值动态聚类方法的可行性和有效性。结果表明,K均值动态聚类法是一种简单有效的方法,可以用于土样的识别和分类,本文方法对工程勘察中的土样分类具有一定的参考价值。
The cluster analysis is a statistical method for grouping the data sets.The K-means cluster technique is applied to the recognition of soil sample in this paper.After describing the basic concepts and calculation procedure of K-means cluster technique,49 soil samples from an engineering investigation field and seven engineering parameters of each sample are chosen to perform K-means dynamic cluster analysis.The sample size effect in the clustering results is analyzed based on the comparison of two clustering results.A clustering result of soil classification using fuzzy C-means cluster analysis method is chosen to verify the feasibility and effectiveness of K-means dynamic clustering.The results show that K-means dynamic clustering is a simple and effective method in classifying soil.The method developed in this paper could be used in the soil’s recognition and classification.And it is referable for classifying soil in the engineering investigation.
出处
《建筑科学》
北大核心
2010年第7期52-56,71,共6页
Building Science
基金
国家自然科学基金(40972191)
上海市教委科研创新项目(09YZ39)
关键词
动态聚类分析
K均值聚类
模式识别
土样
dynamic clustering analysis
K-means cluster analysis
pattern recognition
soil sample
