Reliable stability analysis of surrounding rock for super section tunnel based on digital characteristics of joint information

Abstract

Since all available surrounding rock classification approaches don't consider the size effect of the tunnel excavation span and unfavourable geologic bodies, it's impossible to evaluate the systematic stability of the surrounding rock by rock mass classification alone. In addition, the evaluation results usually have poor robustness because of the uncertainty of structural information. In this paper, the actual distribution information and probability distribution model of the joints in tunnel face are obtained. Many random numbers that meet the probabilistic distribution models can be produced by Monte carlo simulation, and many random rating values of these evaluation indices are also obtained. The distribution probability affiliated with different rock mass levels can be attained by inductive statistics, and a robust evaluation of surrounding rock classification is carried out. The deformation and failure process of surrounding rock are simulated based on the discontinuous deformation analysis. A comparison analysis of removable blocks and their distribution characteristics is carried out by UJN. The most likely damaged parts and rockfall styles are indicated, and a comparison between calculation result and practical collapse of Laohushan tunnel is carried out, which are nearly the same. The research results offer practical guidance for field construction and supporting optimization.

Intelligent recognition of joints and fissures in tunnel faces using an improved mask region‐based convolutional neural network algorithm

“…Leng et al (2021) used the Canny algorithm to detect the edges of joints and fissures in a rock mass and applied fitting, splitting, screening, and merging procedures to the extracted edge detection lines to obtain the boundary lines of the joints and fissures, resulting in a good segmentation effect. However, conventional image processing methods heavily rely on the gray-level difference between joints and fissures and the surround-ing background, making it challenging to be widely used in joint and fissure detection tasks on tunnel faces in complex environments (P. He et al, 2020).…”

“…Other researchers (Sun et al., 2018) used binocular cameras to capture images of tunnel faces and employed techniques such as pixel matching, three‐dimensional coordinate transformation, and image synthesis of overlapping areas to construct a 3D model. CAE Sirovison software (P. He et al., 2020) was then used to recognize structural planes. The above 3D point cloud processing methods can accurately recognize and segment joints and fissures of tunnel faces.…”