CPT based soil behavior classification system

Abstract

Cone penetration test (CPT) has been routinely applied in field investigation owing to its ability to describe soil behavior (SBT) in practice accurately. Besides, CPT results are also used to classify soil behavior with high reliability. The increase in soil data requires regular updating of soil behavior classification systems. In addition to traditional methods, this article applies unsupervised learning algorithms to improve the ability to identify soil groups with similar behavioral characteristics based on a data set including 1339 CPTs with various soil types.

Keywords: CPT; soil behaviors; Unsupervised learning.