Landslide Prediction and Analysis based on Artificial intelligence Approaches in Comparison with Traditional Methods

Description :

This research focuses on developing advanced Artificial Intelligence (AI) models, including Machine Learning (ML) and Deep Learning (DL), to enhance slope stability analysis in geotechnical engineering. A comprehensive dataset of 1,000 cases, consisting of real slope data from the literature and simulated cases generated using Plaxis 2D (FEM) and GeoStudio SLOPE/W (LEM), is utilized to predict the Factor of Safety (FOS) of homogeneous soil slopes. Various ML techniques, such as Support Vector Machines (SVM), Random Forest (RF), Gaussian Process Regression (GPR), and Gradient Boosting (XGBoost), are implemented alongside DL models like Artificial Neural Networks (ANN). Optimization techniques, including classical methods (e.g., ADAM, Mini-Batch Gradient Descent) and nature-inspired methods (e.g., Genetic Algorithm, Particle Swarm Optimization), are integrated to enhance model accuracy and efficiency. The thesis also incorporates Physics-Informed Neural Networks (PINNs) to embed physical laws into model training, ensuring theoretical consistency. By systematically comparing the performance of these AI models with traditional methods, the research validates their reliability and applicability through real-world case studies. This innovative approach addresses the limitations of conventional techniques, offering accurate, efficient, and robust solutions for complex slope stability problems.