Answers are grounded in this paper and cite their sources.
Computational modeling reveals superior parameter inversion using BES-ELM for rock creep in tunnel fault zones, suggesting improved time-dependent stability evaluations.
To determine the creep parameters of surrounding rock in the fault fracture zone of a super-large-section tunnel, the admissible ranges of the Burgers-model creep parameters were first established, after which a FLAC3D numerical model was developed to simulate the construction process of a super-large-section tunnel crossing the fault fracture zone. Fifty learning samples were generated through an orthogonal experimental design, and the corresponding surrounding-rock displacements at the monitoring section were obtained by numerical analysis. Sensitivity analysis was then performed to rank the sensitivities of the creep parameters. Comparative evaluation of Bald Eagle Search-Extreme Learning Machine (BES-ELM), Grey Wolf Optimizer-Extreme Learning Machine (GWO-ELM), and Sparrow Search Algorithm-Extreme Learning Machine (SSA-ELM) indicated that although GWO showed strong performance in fitness optimization, its generalization ability on the prediction set was relatively weak, whereas SSA was competitive in predicting certain parameters. Considering both the parameter-level evaluation indices and the forward-validation results of monitored-section displacements, BES-ELM showed the most favorable case-specific consistency among the three compared models under the present dataset, monitoring section, and 30-day validation period. The creep parameters inverted by BES-ELM can be used as a recommended parameter set for time-dependent stability evaluation at the investigated section and may provide an initial reference for adjacent sections within the same fault-fracture zone under comparable geological and construction conditions.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: