A Simple Direct Method for Prediction of Safety Factor of Homogeneous Finite Slopes

Abstract

This study investigates earthen slope stability problems using numerical simulation with limit equilibrium method. For this purpose 250 dataset is generated via Slope/w software in 2D. The generated dataset is used to develop both a regression and artificial neural network (ANN) models. The dependent variables include: the unit weight of soil ( $$\gamma$$ ), soil cohesion (C), soil friction angle ( $$\varphi$$ ), slope of the embankment with horizontal ( $$\beta$$ ) and the height of slope (H). The independent variable comprises the safety factor of the slope against sliding (Fs). A contour set is produced for assessment of the stability of slopes. The performance of the ANN model in terms of R2 and RMSE is assessed 0.99 and 0.08 respectively and shows superior of the ANN in comparison with the classic regression method. Using the derived regression and ANN models, disregard the need for complicated soft wares implementation. This study produces charts that eliminate the necessity for iteration for Fs in simple homogeneous soil slopes.