Accurately predicting the permeability of coarse-grained soils is crucial for ensuring geotechnical safety and performance. In this study, 64 coarse-grained soil (CGS) samples were designed using a negative exponential gradation equation (NEGE), and computational fluid dynamics–discrete element method (CFD-DEM) coupled seepage simulations were conducted to generate a permeability coefficient (k) dataset comprising 256 entries under varying porosity and gradation conditions. Three machine learning models—a neural network model (BPNN), a regression model (GPR), and a tree-based model (RF)—were employed to predict k, with hyperparameters optimized via particle swarm optimization (PSO) and four-fold cross-validation applied to improve generalization. Gray relational analysis (GRA) revealed that all input parameters (α, β, d_max, n) significantly influence k (R > 0.6). The interquartile range (IQR) method confirmed data suitability for modeling. Among the models, BPNN achieved the best performance (R² = 0.99, MAE = 1.5, RMSE = 2.9, U₉₅ = 0.4), effectively capturing the complex nonlinear relationship between gradation and permeability. GPR (R² = 0.92) was hindered by kernel selection and noise sensitivity, while RF (R² = 0.97) was limited by its discrete regression nature. Compared to a traditional empirical model (R² = 0.9031), BPNN improved prediction accuracy by 10.13%, demonstrating the advantage of data-driven methods for evaluating CGS permeability.

The mechanical characteristics of soil–rock mixtures (SRMs) are significantly influenced by its material composition, composition ratio, water content, and a variety of other influencing factors. A total of 120 SRMs samples with different stone contents and water contents, sourced from typical cohesive soil and crushed stone in the Wuyishan region of Fujian Province, were prepared and subjected to large-scale direct shear tests. This research investigated how stone content, water content, and normal stress impact the shear stress–shear displacement behavior of the SRMs, as well as their influence on internal friction angle, cohesion, and shear strength. The results indicated that the shear stress–shear displacement curves of the SRMs exhibit similar patterns across different conditions, with shear stress increasing with shear displacement. The increase rate correlated closely with water and rock content, as well as normal stress. The shear strength, which adheres to the Mohr–Coulomb criterion, increased with rising stone content. It initially increased and then decreased as water content rose and reached a peak at the optimal water content. Higher stone content reduced shear strength sensitivity to water content changes. The internal friction angle diminished in response to elevated water content, while conversely, it experienced an augmentation with a heightened concentration of stone. It was less affected by water content at higher stone contents. The cohesion decreased as rock content increased, and it initially increased before decreasing with rising water content. Furthermore, as stone content grew, the effect of water content on cohesion became less pronounced.

Taking the shield construction of Xiamen Metro Line 2 tunnel side-crossing the Tianzhushan overpass and under-crossing the Shen-Hai Expressway as the engineering background, FLAC3D 6.0 software was used to examine the deformation of adjacent structures based on shield construction parameters in upper-soft and lower-hard strata. The reliability of the numerical simulation results was verified by comparing measured and predicted deformations. The study results indicate that deformation of the pile will occur during the construction of the tunnel shield next to the pile foundation. The shape of the pile deformation curve in the horizontal direction is significantly influenced by the distance from the pile foundation to the adjacent tunnel’s centerline, as well as by soil bin pressure, grouting layer thickness, and stress release coefficient. During the tunnel shield construction beneath the expressway, increasing the soil bin pressure, the grouting layer thickness, and reducing the stress release coefficient can effectively minimize surface deformation and differential settlement on both sides of the deformation joints between the bridge and the roadbed. The practice shows that, by optimizing shield construction parameters in upper-soft and lower-hard strata, the deformation of nearby bridges and pavements can be kept within allowable limits. This is significant for reducing construction time and costs. The findings offer useful references for similar projects.