基于投影寻踪动态聚类模型的边坡稳定性评价

Slope Stability Assessment Based on Projection Pursuit Dynamic Cluster Model

投影寻踪聚类(PPC)模型在多因素聚类分析中被广泛应用并取得了满意的效果,然而,该模型还存在诸如密度窗宽参数取值经验确定等不足,有待改进提高。本文针对PPC模型存在的问题,把投影寻踪聚类的思想和动态聚类方法结合起来构造投影指标,建立了投影寻踪动态聚类(PPDC)新模型,新模型在整个运算过程中毋需人为给定参数,聚类结果客观、明确。边坡稳定性评价的实际应用表明,PPDC模型不仅切实可行,取得了理想的效果,而且还具有稳定性好、操作简便等特点。PPDC模型为多因素聚类分析提供了一种新方法,具有广阔的应用前景。

The projection pursuit cluster （PPC） model is very popular in application of muiti-factor cluster analysis with satisfactory results. While it still has shortage such as inadequate experience to valuate the parameter of density bandwidth. To solve the problem, in this paper, the ideas of PPC and dynamic cluster method were combined together to set up a new model, the projection pursuit dynamic cluster （PPDC）. During the whole calculation process of the new model, no factitious evaluation parameters were needed. The clustering result was objective and clear. A case of applieation for assessment of slope stability showed that the PPDC model is not only feasible and effective, but also stable and easy to operate in practice. PPDC model provided a new method for multifactor cluster analysis, and it would have a good prospect of wide application.