Acquisition of Rock Mass Mechanical Parameters and Analysis of Surrounding Rock Deformation Characteristics in Jointed Rock Mass Tunnels

ABSTRACT The complexity of joint structures creates substantial uncertainty in determining the mechanical parameters of tunnel surrounding rock, while traditional experimental methods fail to adequately represent actual engineering conditions. To improve the accuracy of rock mass mechanical parameter identification and strengthen the scientific foundation of support design, this study performs a statistical analysis of working face joints using the CAE Sirovision three‐dimensional rock surface scanning system. The combined discrete fracture network (DFN) and discrete element method (DEM) are used to reconstruct the joint distribution network, thereby establishing a high‐precision numerical analysis framework. Based on this framework, a BTO‐BiTCN‐BiGRU‐Attention hybrid inversion model for determining joint mechanical parameters is proposed. The inversion results are then incorporated into DFN–DEM simulations to derive the mechanical parameters of the rock mass and evaluate the deformation characteristics of the surrounding rock. The results indicate that the proposed inversion model outperforms traditional neural network and ensemble learning approaches in prediction accuracy, with a displacement error between calculated and measured values remaining below 7%. The integration of accurately characterized joint distributions and mechanical parameters offers robust data support for tunnel support design and strengthens its scientific rationale.