Surrogate Model Development for Slope Stability Analysis Using Machine Learning

In many countries, slope failure is a complex natural issue that can result in serious hazards, such as landslide dams. It associated with the challenge of stability evaluation, which involves classification problem slopes and regression predicting factor safety (FOS) value. This study explored implementation machine learning to analyze using comprehensive database 880 homogenous (266 unstable 614 stable) based on simulation model developed surrogate model. A was categorize into three classes, including S (stable, FOS > 1.2), M (marginally stable, 1.0 ≤ U (unstable, < 1.0), used predict target The results confirmed efficiency via testing, achieving an accuracy 0.9222, 96.2% for class, 55% 95.2% class. When are same class (i.e., + class), test 0.9315, 93.3% 92.9% low level led minor inaccuracies, be attributed data imbalance. Additionally, found have high correlation coefficient R-square value 0.9989 mean squared error 5.03 × 10−4, indicates strong relationship between values selected parameters. significant difference elapsed time traditional method analysis highlights potential benefits learning.