The Development of the fragility curve for railway embankment.

For the construction of railway embankments, geotechnical engineers pay special attention to slope stability studies. The factor of safety values plays a crucial part in assessing the safe design of slopes. These values determine how close or far slopes are from failing due to natural or man-made causes. While the factor of safety is a numeric indicator of relative stability, it does not indicate the actual risk level of any structure. However, the reliability index and probability of failure quantify the risk level. The present study discusses the findings of a study to determine the factor of safety of an embankment with 12.3 m height using Geo-studio 2012 software. In this article, the fragility curve for six different types of cross-sections was also developed, i.e. the graph between the probability of failure (Pf) and horizontal seismic coefficient (Kh), for various values of Kh (i.e. 0.1, 0.12, 0.144, 0.18, 0.2, 0.3, 0.4 and 0.5). It is observed from the developed fragility curve that as the Kh value increases, the Pf value decreases. A fragility curve can be used to calculate failure probability over a range of seismic zones, and for design purposes, a given seismic zone and probability of failure which determine a unique reliable side slope are selected. Furthermore, two machine learning (ML) models, namely Deep Neural Network (DNN) and Support Vector Regression (SVR) have been developed for the prediction of the factor of safety for different sides slope. The obtained correlation values (R) for SVR and DNN are approximately 0.95 and 0.82, respectively. With the help of the predicted factor of safety, the fragility curve against horizontal seismic coefficient are drawn for both SVR and DNN models. This aims to reduce the time of calculation and facilitate working by suggesting the best model for further analysis of railway embankment. [ABSTRACT FROM AUTHOR]