When performing the inspection of subway tunnels, there is an immense amount of data to be collected and the time available for inspection is short; however, the requirement for inspection accuracy is high. In this study, a mobile laser scanning system (MLSS) was used for the inspection of subway tunnels, and the key technology of the positioning and orientation system (POS) was investigated. We utilized the inertial measurement unit (IMU) and the odometer as the core sensors of the POS. The initial attitude of the MLSS was obtained by using a static initial alignment method. Considering that there is no global navigation satellite system (GNSS) signal in a subway, the forward and backward dead reckoning (DR) algorithm was used to calculate the positions and attitudes of the MLSS from any starting point in two directions. While the MLSS passed by the control points distributed on both sides of the track, the local coordinates of the control points were transmitted to the center of the MLSS by using the ranging information of the laser scanner. Then, a four-parameter transformation method was used to correct the error of the POS and transform the 3-D state information of the MLSS from a navigation coordinate system (NCS) to a local coordinate system (LCS). This method can completely eliminate a MLSS’s dependence on GNSS signals, and the obtained positioning and attitude information can be used for point cloud data fusion to directly obtain the coordinates in the LCS. In a tunnel of the Beijing–Zhangjiakou high-speed railway, when the distance interval of the control points used for correction was 120 m, the accuracy of the 3-D coordinates of the point clouds was 8 mm, and the experiment also showed that it takes less than 4 h to complete all the inspection work for a 5–6 km long tunnel. Further, the results from the inspection work of Wuhan subway lines showed that when the distance intervals of the control points used for correction were 60 m, 120 m, 240 m, and 480 m, the accuracies of the 3-D coordinates of the point clouds in the local coordinate system were 4 mm, 6 mm, 7 mm, and 8 mm, respectively.

The expansion of tunnel scale has led to a massive demand for inspections. Light Detection And Ranging (LiDAR) has been widely applied in tunnel structure inspections due to its fast data acquisition speed and strong environmental adaptability. However, raw tunnel point-cloud data contain noise point clouds, such as non-structural facilities, which affect the detection of tunnel lining structures. Methods such as point-cloud filtering and machine learning have been applied to tunnel point-cloud denoising, but these methods usually require a lot of manual data preprocessing. In order to directly denoise the tunnel point cloud without preprocessing, this study proposes a comprehensive processing method for cross-section fitting and point-cloud denoising in subway shield tunnels based on the Huber loss function. The proposed method is compared with classical fitting denoising methods such as the least-squares method and random sample consensus (RANSAC). This study is experimentally verified with 40 m long shield-tunnel point-cloud data. Experimental results show that the method proposed in this study can more accurately fit the geometric parameters of the tunnel lining structure and denoise the point-cloud data, achieving a better denoising effect. Meanwhile, since coordinate system transformations are required during the point-cloud denoising process to handle the data, manual rotations of the coordinate system can introduce errors. This study simultaneously combines the Huber loss function with principal component analysis (PCA) and proposes a three-dimensional spatial coordinate system transformation method for tunnel point-cloud data based on the characteristics of data distribution.

This study investigates the micro-mechanical behavior of mortar under uniaxial compression using a three-phase model in PFC^3D. By simulating mortar as a composite of cement, sand, and the interfacial transition zone (ITZ), the research examines the impact of particle size on stress–strain behavior, crack propagation, porosity distribution, contact forces, and energy transformation. The simulations reveal that reducing sand particle size from 1–2 mm to 0.25–0.5 mm leads to a significant increase in uniaxial compressive strength, with peak strength values rising from 65.3 MPa to 89.6 MPa. The elastic modulus similarly improves by approximately 20% as particle size decreases. The study also finds that tensile cracks dominate failure, accounting for over 95% of total cracks, with their onset occurring at lower strains as the particle size is reduced. Porosity analysis shows that smaller particles result in a more uniform distribution, with the final porosity at peak strength ranging between 0.26 and 0.29, compared to 0.22 to 0.31 for larger particles. Additionally, energy dissipation patterns reveal that as particle size decreases, the boundary energy transformation into strain energy becomes more efficient, with a 15% increase in strain energy storage observed. These findings provide critical insights into optimizing mortar microstructure for enhanced mechanical performance in construction applications.

The acquisition of tunnel inspection data is fundamental to tunnel operation and maintenance. Existing inspection equipment is either based on laser scanning or CCD cameras. Laser scanning-based devices often struggle to detect cracks, while equipment based on CCD cameras is unable to acquire point cloud to assess tunnel deformations. In order to obtain high-quality, comprehensive data, this paper develops a mobile three-dimensional inspection device CKY-200, integrated with multiple sensors. The CKY-200 addresses the temporal and spatial synchronization issues of CCD cameras and laser scanners. Considering the convenience of inspection, the device progresses on the track powered by electricity. Moreover. This article proposes a tunnel deformation detection algorithm and a crack width measurement method, and verifies the accuracy of the data collected by the equipment. Through on-site experiments and comparison, the CKY-200 provides the advantages of more comprehensive data acquisition while ensuring high precision and speed.

To address the challenges of irregular geometries, multi-scale variation, and diverse target types in road damage detection, we propose an improved deep learning model based on YOLOv8, named YOLO-DMD (Deform–Multiple–Dynamic). The model enhances detection performance through three key components: C2f-Deform, which introduces deformable convolutions to improve adaptability to irregular geometric features; a multi-scale dilated attention module that integrates multi-scale dilated convolutions with attention for enhanced feature representation; and a dynamic head that decouples scale-, spatial-, and task-aware representations to refine classification and localization. Experiments on the public SVRDD dataset show that YOLO-DMD achieves an mAP@0.5 of 79.8% and a recall of 73.3%, demonstrating competitive and often superior performance compared with several mainstream detectors, including YOLOv5 and YOLOv8. Ablation studies further verify the individual effectiveness of each module and their complementary interactions. In addition, cross-dataset evaluations on RDD2020 and GRDDC-US indicate improved robustness under domain shift. Although YOLO-DMD introduces additional computational parameters, it maintains real-time inference capability and provides a practical balance between accuracy and efficiency, making it suitable for automated road damage detection.