Accurate evaluation of cumulative strains in marine soils under long-term cyclic loading is essential for the design and safe operation of offshore wind turbines. This study proposes an enhanced machine learning model to predict the cumulative strain in marine soils subjected to cyclic loading. Cumulative strains of marine soils from five offshore wind farms under long-term cyclic loading were tested. Four prediction models for cumulative strains were developed and evaluated based on test results using the Back Propagation Neural Network (BP-NN), Random Forest (RF), Support Vector Regression (SVR), and eXtreme Gradient Boosting (XGBoost) models, each combined with the Particle Swarm Optimization (PSO) algorithm. The prediction model with the highest accuracy was further analyzed using the SHapley Additive exPlanations (SHAP) method. Results show that the RF and XGBoost algorithms have higher prediction accuracy, with R2 values above 0.99, compared to the BP-NN and SVR models. Furthermore, dynamic triaxial test parameters significantly influence the cumulative strain predictions more than the soil properties. This study provides a more efficient method for cumulative strain assessment of marine soils under long-term cyclic loading.

The utilization of cement has been found to have negative environmental impacts. In order to reduce the quantity of cement used and improve the mechanical properties of solid waste-cement-stabilized cohesive soil, the incorporation of solid waste as additives has been investigated. Unconfined compressive strength is a crucial parameter in geotechnical engineering. However, existing empirical formulas have limited accuracy and applicability when it comes to the unconfined compressive strength of solid waste-cement-stabilized cohesive soil. The machine learning model can be used to provide accurate and comprehensive predictions by considering the nonlinear relationships between independent and dependent variables. This study aims to propose a machine learning model tuned by optimization algorithms with high generalization performance in accurately predicting the unconfined compressive strength. Firstly, a database containing 474 specimens was developed. Secondly, eight machine learning models were established, composed five single models and three hybrid models, to train and test the database. Six performance indicators were employed to evaluate the generalization ability of these models. Finally, the optimal model was selected for analysis of the importance of the feature variables using shapley additive explanations, which were compared with those of the existing empirical model. The research findings indicated that, the extreme gradient boosting model tuned with tree-structured parzen estimators exhibited the highest predictive accuracy and generalization ability. The curing age, cement content, plastic limit, and water content were identified as the most critical factors influencing the unconfined compressive strength. Among the chemical components in solid waste, the aluminum oxide content and silicon dioxide content were found to significantly influence the unconfined compressive strength, while the impact of calcium oxide content was relatively minor. Furthermore, the optimal solid waste content was found to be around 10 %. This study made a significant contribution to the effective utilization of waste resources in the context of sustainable construction practices.

Despite the widespread application of data-centric techniques in Geotechnical Engineering, there is a rising need for building trust in the artificial intelligence (AI)-driven safety assessment of road embankments due to its so-called black-boxnature. In addition, from the lens of limit equilibrium approaches, e.g., Bishop, Fellenius, Janbu and Morgenstern-Price, and finite element method, it is essential to carefully examine the interplay of both topological and physical/mechanical properties during the safety factor (FoS) predictions. First, aside from having conventional geotechnical inputs for soil in core and foundation and the height of embankments, this paper codifies geometric features innovatively. The number of slope types with different ratios including 1:1, 1.5:1 and 2:1 as well as the number of berms is introduced. Second, a pool of 19 machine learning (ML) techniques is effortlessly trained on the dataset using an automated ML (AutoML) pipeline to identify the most optimized ML algorithm. Finally, to achieve post-hoc interpretability for the internal mechanism of the input- output relationship unbiasedly, a game-theory-based explainable AI (XAI) method called Shapley additive explanations (SHAP) values is applied. SHAP-aided importance analysis provides human-interpretable insights and indicates height, California bearing ratio, slope type 2:1 and cohesion as the most influential parameters. Exclusively, analyzing hazardous embankments by classifying main and joint contributors exhibits a complex and highly variable influence on the FoS. This paper harnesses the power of XAI tools to enhance reliability and transparency for the rapid FoS prediction of slopes. It targets geotechnical researchers, practitioners, decision-makers, and the general public for the first time.