Multiple Parameters Determination Method of Hardening Soil Model Based on Particle Swarm Optimization

The utilization of numerical analytic techniques has become essential in evaluating the environmental consequences of excavation in foundation pits. The careful selection of suitable soil constitutive models has significant importance in this regard. The Hardening Soil (HS) model has become widely utilized in the numerical analysis of foundation pits among the several models considered. The essence of a constitutive model is in the identification and determination of its calculation parameters. The stress-strain curve is derived by conducting triaxial testing, utilizing the inherent properties of the particle swarm optimization (PSO) technique. The objective function entails quantifying the disparity between stress-strain curves obtained from various parameters and experimental curves while iteratively searching for five crucial parameters of soil-hardening models: effective internal friction angle phi, effective cohesion, failure ratio Rf, stiffness level-dependent power exponent, and standard triaxial-drained test secant stiffness E50ref. The objective of this research is to create a computational software utilizing the Python programming language to execute PSO with the purpose of determining sets of five fundamental soil constitutive parameters. This research employs optimized parameters acquired from PSO to construct a numerical model for foundation pits and performs numerical computations. Accurate findings may be reached by comparing the measured deformation in the numerical simulation with that seen in actual foundation pits. The results demonstrate a high degree of consistency between the experimental curves and the five parameter combinations obtained through the PSO algorithm, with an error rate not exceeding 8.2%. Moreover, the optimized curves exhibit closer alignment with theoretical expectations. The settlement values calculated using a numerical model based on optimized parameters show only a 6.12% deviation from actual measurements while closely following the trend of the observed curve. This outcome has the potential to greatly alleviate the burden of conducting indoor tests and holds considerable practical value in engineering applications. The observable optimization impact of PSO also serves as evidence of its viability in improving soil constitutive parameters, providing new perspectives and sources of information for future progress in this domain.