论文部分内容阅读
在岩体斜坡稳定性分析和岩体水力学分析中,岩体随机结构面的优势分组是一项十分重要的内容。提出一种基于凝聚层次聚类分析的岩体随机结构面产状优势分组的新方法,这种方法的优点在于事先无需确定聚类中心,在分类结果生成后还可明显剔除数据的孤点与野值。应用人工随机生成的结构面产状数据对这种新方法和模糊C均值法进行了对比验证。结果表明,凝聚层次聚类分析法不仅在无孤值点的情况下分组结果优于模糊C均值算法,而且还可以有效地剔除孤值点对于分组结果的不利影响。最后将这种方法应用于松塔水电站坝肩结构面优势分组中,同样得到了比较满意的结果。
In rock slope stability analysis and rock mass hydraulic analysis, the superior group of rock mass random surface is a very important content. A new method based on cohesive hierarchical clustering analysis is presented, which has the advantage of generating grouping of stochastic structural planes in rock masses. The advantage of this method is that it does not need to determine the cluster centers in advance, and can obviously eliminate the isolated points and fields of data after the classification results are generated value. This new method and the fuzzy C-means method are compared and validated by artificial randomly generated surface data. The results show that the cohesive hierarchical cluster analysis method not only has better clustering results than fuzzy C-means algorithm but also has no effect on the grouping results. Finally, this method is applied to the superior group of dam abutment structure of Songta Hydropower Station, and the satisfactory results are also obtained.