Application of AI models for reliability assessment of 3d slope stability of a railway embankment

• Published in: Multiscale and Multidisciplinary Modeling, Experiments and Design Vol. 7; no. 2; pp. 1007 - 1029
• Main Authors: Rao, Brijbhan, Burman, Avijit, Roy, Lal Bahadur
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 01.06.2024
• Subjects: Characterization and Evaluation of Materials; Engineering; Mathematical Applications in the Physical Sciences; Mechanical Engineering; Numerical and Computational Physics; Original Paper; Simulation; Solid Mechanics; 3D slope stability; Pore-water pressure; Bishop’s method; MARS; LSSVM; SVM; Seismic loading
• ISSN: 2520-8160; 2520-8179
• DOI: 10.1007/s41939-023-00255-9

In this work, three machine learning (ML) techniques i.e., multivariate adaptive regression splines (MARS), support vector machine (SVM), and least square vector machine (LSSVM) have been used to conduct a reliability analysis of a railway embankment situated in the Mokama district of the state of Bihar, India based on 3D slope failure investigation. The probabilistic assessment of the railway embankment is important as it incorporates risk estimates of a real-life structure against various uncertain loading conditions. The main goal of the analysis is to evaluate the stability of the embankment in three dimensions while taking pore pressure and seismic impacts into consideration. Using Scoops-3D software, the Factor of Safety (FOS) of the embankment is calculated based on Bishop’s Simplified Method. Notably, to determine the reliability index ( β ), the study takes uncertainties resulting from geographical differences in soil parameters into account. Our findings show that the railway embankment functions admirably under various loading conditions. For the training and testing datasets, error markers such as root mean squared error (RMSE), Weighted Mean Absolute Percentage Error (WMAPE), Nash–Sutcliffe coefficient (NS), Performance index (PI), Maximum determination coefficient value ( R 2 ), Adjusted R 2 , Scatter Index (SI), Uncertainty at 95% (U 95 ), Objective function (OBJ) criterion values are determined to assess the models’ capacity for prediction. By proving the effectiveness of machine learning approaches in capturing complex interactions and uncertainties, this research advances the subject of railway embankment stability evaluation. Among the evaluated models, the LSSVM model proved to be the best accurate predictor.