Reliability Analysis of Seismic Foundation Bearing Capacity via Machine Learning and Monte Carlo Simulation

The assessment of seismic foundation bearing capacity is crucial for ensuring the stability and safety of buildings, hydraulic structures, and bridges. Traditional safety factor methods are limited in capturing parameter uncertainties and quantifying failure probabilities. To address this issue, This study develops an efficient reliability assessment framework by integrating FELA with a Multi-Layer Perceptron (MLP) model. The FELA method is employed to compute seismic foundation bearing capacity and generate a large-scale dataset encompassing diverse parameter combinations. This dataset is used to train the MLP model, which establishes a highly accurate mapping between input parameters and ultimate bearing capacity. Monte Carlo Simulation (MCS) is incorporated to quantify failure probabilities and reliability indices under static and seismic conditions, significantly improving computational efficiency. The analysis determines a reliability index threshold of β = 2.58 for seismic conditions, providing a probabilistic reference for foundation safety evaluation. An engineering case study validates the effectiveness and practical applicability of the proposed approach. The combination of FELA and machine learning offers a robust and accurate method for seismic foundation reliability assessment, advancing the optimization of seismic foundation design.