Study on the Creep Characteristics of Gentle Dip Red Bed Mudstone of the Surrounding Rock Tunnel

Red bed soft rock exhibits poor physical and mechanical properties and certain creep characteristics, which cause continuous deformation of tunnels and underground engineering structures during operation. The rock mass at the bottom of a railway tunnel in Sichuan Province is mudstone of the Penglaizhen Formation of the Upper Jurassic. It has a thin to medium-thick layered structure. Moreover, the rock formation is nearly horizontal. In this paper, the shear creep test and deep learning are employed to study the creep characteristics of gentle dip red bed mudstone.The results indicate that the red bed mudstone in the tunnel site exhibits medium-low creep characteristics. When the stress level is relatively low, the rate of the creep deformation gradually reduces with time, and when it reaches a certain time, the deformation no longer increases, and the final deformation tends to a stable value. Conversely, when the stress level is relatively high, although the rate of the creep deformation gradually reduces as the time increases, it remains unchanged when it reduces to a certain value. At the initial stage of creep, the deformation of each grade is evident; at the middle stage, the deformation is slow; and at the later stage, the deformation remains unchanged, that is, it enters the stable creep stage after a rapid decay in creep rate. The long-term strength of red bed mudstone in the tunnel site is low, which easily causes continuous deformation of the surrounding rock of tunnels under the action of high ground stress. According to the long-term deformation monitoring data at the bottom of the tunnel, combined with 20 groups of red bed mudstone creep parameter samples and the upper arch deformation data of the tunnel numerical model, the deep learning inverse analysis model of the creep parameter of rock mass is established based on the deep learning algorithm. Finally, we obtain the creep parameter of the red bed mudstone via inverse analysis. The research results provide a basis for engineering structure design and long-term deformation prediction of red bed mudstone strata in this area.