Facing the needs of deep-water and large-scale port construction, this paper proposes a multi-algorithm dynamic coupling optimization framework of "FEM-BEM-ML" to solve the problems of insufficient accuracy and low efficiency of traditional single algorithm in simulating the dynamic response of port channel structure. Finite element method (FEM) is used in solid domain to capture material nonlinearity and pile-soil interaction with high accuracy. Boundary element method (BEM) is used in fluid domain to efficiently solve wave radiation and diffraction in infinite domain. Adaptive coupling criterion based on local acceleration error estimation is introduced into the interface to realize dynamic adjustment of step size under time-varying load. Then, the proxy model is constructed by Gaussian process regression (GPR) instead of time-consuming full coupling calculation, and the key parameters such as pile diameter and wall thickness are optimized by coupling genetic algorithm (GA) to achieve the balance between structural safety and material consumption. The typical deep-water piled wharf case shows that the stress prediction error of this method is reduced from 12.5% to < 1% compared with the traditional Morison equation method, and the calculation time is reduced by 3 orders of magnitude. After optimization, the volume of single pile concrete is saved by 8%, and the stress and displacement meet the specifications. The research provides a reliable technical way for the refined and efficient design of port channel structure in complex marine environment.
Cui et al. (Sun,) studied this question.