Synergizing FEM with AI: Leveraging Dense Neural Networks, Random Forests, and SHAP for Enhanced Feature Importance Analysis in Long-Term Arch Dam Displacement Prediction

This study combines advanced machine learning techniques with finite element modeling to analyze the structural behavior and displacement of a super-high concrete arch dam. By integrating finite element analysis, dense neural networks, and random forest approaches, the research determines the relative importance of factors influencing dam displacement over long-term operations. The study utilizes SHAP (SHapley Additive exPlanations) values to interpret the results, providing deeper insights into how each factor contributes to the model's predictions. The findings highlight lake level as the most significant factor, followed by thermal effects and creep, demonstrating the potential of this integrated approach for enhancing structural health monitoring of large infrastructure projects.