A systematic review and comparative assessment of optimization algorithms in geotechnical engineering with a benchmark foundation design case study

Engineering notes

To augment the literature synthesis and ensure methodological soundness, a benchmark case study on shallow foundation optimization is presented. Although all methods converge to similar optimal designs, the surrogate-based GPR-BO achieves equivalent solution quality with significantly fewer function evaluations and improved computational efficiency, while also providing probabilistic predictions and uncertainty quantification.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Optimization algorithms are becoming more prevalent in geotechnical engineering, particularly for tackling challenges involving nonlinearity, uncertainty, and intricate soil–structure interactions. Despite this, most current review studies are primarily descriptive and do not provide a systematic, criteria-driven comparison of optimization methodologies. This study presents a systematic review and a structured comparative evaluation of classical, metaheuristic, surrogate-based, and artificial intelligence–based optimization techniques used in geotechnical engineering. The methods under review are assessed using a unified framework that considers convergence efficiency, data efficiency, interpretability, uncertainty quantification capabilities, constraint handling, scalability, and practical maturity. To augment the literature synthesis and ensure methodological soundness, a benchmark case study on shallow foundation optimization is presented. Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Gaussian Process Regression–based Bayesian Optimization (GPR-BO) are employed to address an identical constrained footing design problem subject to settlement and bearing capacity constraints. Although all methods converge to similar optimal designs, the surrogate-based GPR-BO achieves equivalent solution quality with significantly fewer function evaluations and improved computational efficiency, while also providing probabilistic predictions and uncertainty quantification. This systematic comparison and quantitative case study collectively demonstrate that optimization methods in geotechnical engineering are not interchangeable, underscoring the benefits of data-efficient, uncertainty-aware optimization frameworks for dependable, sustainable geotechnical design.

Links and sources

• PDF from original source

Comments