Deep foundation pit support structures in soft soil are susceptible to deformation and instability under tidal action, making accurate prediction of their mechanical response and reliability assessment crucial for engineering safety. First, a three‐dimensional numerical model incorporating dynamic tidal loads was developed in ABAQUS to simulate the mechanical responses under different tidal conditions, and its accuracy was validated through physical model tests. Second, to overcome the limitation of traditional backpropagation (BP) neural networks converging to local minima, an improved particle swarm optimization (PSO) algorithm was employed to optimize their weights and thresholds. Subsequently, a mechanical response prediction method based on the improved PSO–BP neural network was proposed and verified to exhibit higher accuracy and generalization capability than the standard BP network. Finally, reliability analysis was conducted using the established model. The results indicate that increased tidal intensity leads to more dispersed stress and displacement distributions of the support structures, accompanied by a significant rise in uncertainty. Mitigation measures, including enhancing structural stiffness, optimizing waterproof curtain performance, and installing effective drainage systems, are recommended to reduce the adverse impact of tidal fluctuations on foundation pit stability. This study provides a novel theoretical framework and technical support for the mechanical response prediction and reliability assessment of deep foundation pit support structures in soft soil under tidal action.
Ma et al. (Thu,) studied this question.