摘要
针对传统的岩体结构面分组方法的缺点,提出了基于模糊聚类的结构面动态聚类方法。首先以模糊聚类法对结构面样本进行分组,得到优势分组及各组优势产状,并将其作为动态聚类法的初始划分;再采用动态聚类法对结构面样本进行更进一步的聚类分析。该法克服了动态聚类算法对初始聚心敏感及易收敛于局部极小值的缺点,提高了分组结果和优势产状的客观性及准确性。将算法应用于野外实测岩体结构面产状数据分析中,得到的聚类结果合理、可靠,符合实际情况。
Aiming at the shortcomings of the traditional classifying analysis of rock structural interfaces,a dynamic clustering method based on fuzzy clustering is presented.We firstly use the fuzzy clustering to classify the structural interfaces samples and obtain the prevailed grouping and the prevailed occurrence of the samples,and take the results as the initial partition of dynamic clustering to further group the structural interfaces samples by dynamic clustering method.It can overcome the shortcomings,including sensitivity to initialization and convergence to a local minimum and improve the objectivity and accuracy of analysis result.The presented algorithm has been used in analyzing the field measured occurrence data of discontinuities in real rock mass,and the results is reasonable and credible comparing with the fact.
出处
《人民长江》
北大核心
2012年第9期59-63,共5页
Yangtze River
基金
国家自然科学基金项目(40972195)
关键词
岩体
结构面
模糊聚类
动态聚类
rock mass
structural interface
fuzzy clustering
dynamic clustering
作者简介
宋金龙，男，硕士研究生，主要从事岩体稳定及环境地质工程方面的研究工作。E—mail：boris—song@163．com
