Data and code for: Leakage-Free Machine-Learning Surrogates for Embankment-Induced Ground Settlement and Existing-Tunnel Response: A Synthetic Finite-Element Benchmark
Year: 2026
Type: article
Abstract: Finite-element databases, audit studies, machine-learning results, trained surrogates and scripts behind the manuscript of the same title (Transportation Geotechnics, revision R1). The data are synthetic: two-dimensional plane-strain, drained Kratos Multiphysics 10.4 analyses of a staged embankment over an existing 6 m tunnel in soft clay. Contents: three Latin-hypercube databases of 150 cases x 8 fill stages (seeds 42, 7, 2026) with meshes, stage fields and the canonical stage tables; a free-field reference without the tunnel; mesh, interface, one-at-a-time, load-step, construction-history, layered-modulus and undrained audit studies; case-disjoint fold assignments and out-of-fold predictions of eight regressors for eight targets; 64 surrogates fitted with the fixed hyperparameters of the paper; software versions. README.md gives the column dictionary and the commands that reproduce every table and figure. Licences: CC BY 4.0 (data, results, models), MIT (code).
Source: Open Science Framework
Author: TRAN VU VAN HOA
Language: English
Subfield: Civil and Structural Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225617408
FWCI:
Cites:
Cited by:
Related to: