Just Accepted

This article has been peer reviewed and accepted for publication. It is in production and has not been edited, so may differ from the final published form.

Identification method of dike contact seepage hidden danger based on saturation line characteristics

Quanyi Xie, Hongzheng Luo, Zhanhua Li, YUSEN WANG , Gaohang Lv, Peizhi Zhuang, Lizhuang Cui, Jian Liu

Abstract

Context. The contact seepage between the embankment and the trans-dam culvert is a crucial factor that can lead to embankment failure. Aims. The purpose of this article is to examine an algorithm that can intelligently identify contact seepage diseases based on the characteristics of saturation lines. Methods. This study investigates the variation in unsaturated seepage parameters of soils with different compactness and the changes in saturation lines over time, based on the theory of saturated-unsaturated seepage. Based on the foregoing, the capability of various classification algorithms in identifying a saturation line resulting from contact seepage is evaluated. Key results. To select the saturation line characteristic indicators applicable for identifying embankment seepage, 45 sets of saturation line characteristic indicators were divided into 7 combinations. The feature combination 2 (i.e. the average slope of the time-series data) is the most suitable among the seven combinations for identifying contact seepage of embankments. PSO-ELM has the highest accuracy and F1 score in identifying contact seepage among all algorithms, reaching 99.23% and 99.52 respectively. Conclusions. The PSO-ELM is the most suitable algorithm for distinguishing whether contact seepage occurs. Implications. Machine learning algorithms have the capability to differentiate the health condition of embankments, facilitating non-destructive, intelligent, and swift assessments of embankment health.

SR24002  Accepted 11 August 2024

© CSIRO 2024