多标度数据几何轮廓相似性的度量与应用

Similarity Measure of Geometric Contours about Multi-scale Data and Its Application

在多标度数据的分类问题中,对于分布特征不清或小样本下的数据的相似性度量仍是研究的热点.对此,建立了基于样本几何轮廓相似度的判别分析模型,并应用于采动沉陷建筑物损坏等级的评价,结果表明该模型简洁、有效,无需数据预处理,较少依赖先验信息,具有推广应用的价值.

In the classification problem of multiple scale data, similarity measure between data spaces with unclear distribution characteristics or small sample data is still a hot re- search topic. Therefore, this paper establishes the discriminant analysis model based on the geometric contours similarity of sample data, and we apply it into damage level evaluation for mining subsidence building. Results show that the model is simple and effective, ：without data preprocessing, less dependent on prior information. Thus, the model is worth promoting in practice.