Search

Estimating active earth pressure in cohesionless backfill behind rigid retaining walls has been significantly improved through the application of cutting-edge soft computing techniques in this paper. This study delves into the intricate interplay of negative wall-soil friction with load transfer mechanisms, underpinned by a comprehensive dataset obtained from a conservative solution that leverages statically admissible stress fields. Utilizing a Bayesian regularization backpropagation neural network, we construct an explicit function for active earth pressure estimation, ensuring the model's reliability through its remarkable alignment with measured values. Furthermore, feature importance analysis and advanced mathematical modeling enrich the study, providing a practical tool for retaining wall design and analysis. This tool is adept at addressing complex scenarios, including those involving negative wall-soil friction, thereby advancing the state-of-the-art in geotechnical engineering. [ABSTRACT FROM AUTHOR]