Slope Stability Analysis Using Rf, Gbm, Cart, Bt and Xgboost

Slopes in geotechnical and mining engineering are the most crucial geo-structure. Predicting or forecasting the stability or instability of the slope and then classifying the slope accordingly helps in mitigating the risks and enhancing the design by maximizing the safety. Computing techniques have overpowered the analytical and statistical models used for predicting the stability of the slopes. To reduce the uncertainties and ambiguity of the previously used models, lately, researchers have come up with the novel techniques for Slope Stability Classification (SSC) which are Random Forest, Gradient Boosting Machine, Extreme Gradient Boosting, Boosted Trees and Classification and Regression Trees. These computational algorithms are employed in this research paper and the slope details are taken from a literature i.e. 221 input datasets are used and slopes are classified accordingly using the mentioned models. The relation between the inputs such as height (H), slope angle (beta), cohesion (c), pore water pressure ratio (ru), unit weight (gamma), angle of internal friction (phi) and slope stability (output) is established and slopes are categorized according to their failure and stability. Performance analysis is done thereafter to analyses and compare different models and let the readers and researchers know that which model sufficed and fitted best to the study.

Automated summary

This study explores novel techniques for Slope Stability Classification (SSC) and uses computational algorithms to classify slopes based on 221 input datasets. By establishing the relationship between input parameters, slopes are categorized into different failure and stability classes, followed by performance analysis to compare the suitability of different models.