In the construction of deep foundation pits, early warning measures are essential to reduce construction risks and prevent personnel injuries. In underground structure and pressure analysis, soil layer and support structure data are indispensable. Therefore, soil layer reconstruction serves as a critical step, while sparse borehole data limit the accuracy of traditional reconstruction methods. This paper proposes a machine learning-based soil layer reconstruction method to address this issue. First, various types of borehole and soil layer data are generated by simulating the formation process of Earth’s soil layers, thereby providing sufficient training data. Subsequently, a coding algorithm is designed to extract soil layer features as inputs for the convolutional neural network. Finally, 3D meshing is performed on the soil layer generated from real boreholes, and soil model rendering is achieved through a voxel clustering algorithm. The algorithm achieved an accuracy rate of over 90% in tests and demonstrated excellent robustness. By applying this algorithm, we successfully reconstructed the soil layers at a typical foundation pit site in a Chinese city, validating its effectiveness in real-world scenarios and its potential for large-scale engineering applications.

As a novel solderless vertical interconnection structure, the fuzz button exhibits broad application prospects in high-density electronic packaging and high-reliability interconnection systems owing to its low profile, separable interconnection capability, and excellent environmental adaptability. However, due to its complex internal geometric configuration and contact nonlinearity, conventional equivalent modeling approaches are insufficient to accurately characterize its actual mesoscopic structure and mechanical behavior. To address this issue, a finite element simulation method for fuzz buttons based on virtual fabrication was proposed. A trajectory generation method describing the “movement–direction change–fixation–reset” process of metallic wires was first established based on a three-dimensional braiding technique, and cubic spline interpolation was employed to smooth and fit the trajectories, thereby constructing a three-dimensional preform model of the fuzz button. Subsequently, the compression molding process was implemented in Abaqus to obtain a fuzz button model possessing realistic geometric morphology and mesoscopic contact topology characteristics. Considering the pronounced contact nonlinearity and large deformation behavior, the penalty function method was adopted to handle the complex contact interactions among metallic wires, while an explicit dynamic solver with displacement-controlled loading was employed for numerical computation. The mesh was discretized using C3D8R solid elements, and local mesh refinement was introduced in the contact regions to balance computational accuracy and efficiency. Key mechanical response data, including reaction force–displacement curves and energy dissipation characteristics, were extracted and validated experimentally. The results demonstrate that the proposed finite element simulation method can effectively reproduce the compressive behavior and nonlinear hysteretic response of fuzz buttons, with good agreement achieved between simulation and experimental results. The proposed approach provides an effective tool and theoretical basis for mesoscopic structural modeling, mechanical performance analysis, and structural optimization design of fuzz button contacts.

Fuzz buttons are formed by interweaving and compacting fine metallic wires, resulting in a highly porous architecture with complex internal contact interactions. Their compressive behavior is governed by the evolution of wire–wire contacts, frictional sliding, local bending, and plastic deformation, which cannot be adequately captured by conventional homogenized models. To address this limitation, a process-informed finite element modeling approach based on virtual fabrication is proposed. First, the spatial trajectories of 24 beryllium copper wires are generated using a parametric three-dimensional weaving algorithm and smoothed by cubic spline interpolation to obtain continuous wire centerlines. The resulting preform is then virtually compacted to reconstruct the densified wire network and its contact topology. The model employs a globally controlled solid-element mesh, a penalty-based general contact algorithm, a Coulomb friction model, and an explicit quasi-static solution scheme. The size-dependent plastic response of the fine wires is further incorporated through a Nix–Gao-based correction to the constitutive relation. The model is validated against quasi-static compression experiments at compressive strains of 15%, 20%, and 25%. The relative errors in the predicted peak forces are 2.12%, 5.65%, and 6.81%, respectively, while the corresponding coefficients of determination for the force–displacement curves are 0.984, 0.970, and 0.973. The model successfully reproduces the nonlinear loading–unloading response and hysteretic energy dissipation over the investigated strain range. The proposed approach provides a physically grounded numerical framework for predicting the compressive behavior of fuzz buttons and investigating the mesoscopic mechanics of complex interwoven wire networks.