Kun Zeng

In the face of escalating environmental concerns and the urgent need for sustainable construction materials, this study explores the innovative use of waste red mud, an abundant byproduct of the aluminum production process, in the production of concrete curbstones. The unique physicochemical properties of red mud, including its fine texture, high alkalinity, and rich composition of iron, aluminum, and other oxides, are leveraged to enhance the traditional concrete mix. This research delves into the optimal incorporation of red mud into concrete curbstones, focusing on the mixture ratios, physical properties, and resultant strength. Optimal results were observed with a 10 % replacement ratio of red mud, which improves flexural strength; however, higher proportions of red mud (exceeding 30 %) led to a decrease in strength, indicating a threshold for its effective use in enhancing concrete properties. In addition, the study reveals that while high red mud content may induce surface porosity due to uneven vibration, it does not cause significant structural flaws like cracks or stratification. This study establishes red mud as a promising sustainable alternative for replacing traditional materials in concrete curbstone production, paving the way for more eco-friendly practices in highway engineering and construction.

Gas-insulated transmission lines (GILs) are increasingly deployed in dense urban power networks, where complex construction activities may introduce external mechanical impacts and pose risks to pipeline structural integrity. However, existing GIL monitoring approaches mainly emphasize electrical and gas-state parameters, while lightweight solutions capable of rapidly detecting and localizing impact-induced structural anomalies remain limited. To address this gap, this paper proposes an intelligent end-edge-cloud monitoring system for impact anomaly detection in GIL pipelines. Numerical simulations are first conducted to analyze the dynamic response characteristics of the pipeline under impacts of varying magnitudes, orientations, and locations, revealing the relationship between impact scenarios and vibration mode evolution. An end-tier multi-parameter intelligent sensor is then developed, integrating triaxial acceleration and angular velocity measurement with embedded lightweight computing. Laboratory impact experiments are performed to acquire sensor data, which are used to train and validate a multi-class extreme gradient boosting (XGBoost) model deployed at the edge tier for accurate impact-location identification. Results show that, even with a single sensor positioned at the pipeline midpoint, fusing acceleration and angular velocity features enables reliable discrimination of impact regions. Finally, a lightweight cloud platform is implemented for visualizing structural responses and environmental parameters with downsampled edge-side data. The proposed system achieves rapid sensor-level anomaly detection, precise edge-level localization, and unified cloud-level monitoring, offering a low-cost and easily deployable solution for GIL structural health assessment.

Ensuring the quality and safety of large‐diameter tunnels requires accurate, real‐time deformation monitoring. However, existing methods often lack the precision needed for larger shield tunnel structures, are difficult to implement during construction, and rarely provide real‐time data. This study presents a novel approach for estimating the deformation profile of large‐diameter shield tunnels using sensor technology. Sensors are installed immediately after tunnel segment assembly to monitor key parameters, specifically segment rotation and seam openings. A finite element method (FEM) model establishes a theoretical relationship between these measurements and the overall tunnel profile, allowing the actual deformation to be estimated based on real‐time sensor data. To enhance the robustness of the results and reduce errors from individual sensors, a Kalman filter is employed to fuse the two estimated profiles. The study’s findings indicate that deformations estimated from segment rotation were less severe than those derived from seam openings. The Kalman filter–based fusion reduced the influence of individual sensor‐derived errors and provided more stable tunnel profile estimates. Compared with total‐station measurements, the fused profiles achieved MAE values of 2.07–3.30 mm and RMSE values of 2.70–4.16 mm during the validation period, with average MAE and RMSE of 2.70 and 3.41 mm, respectively. These results demonstrate the feasibility of the proposed framework for real‐time deformation monitoring of large‐diameter shield tunnels.

Detecting cavity‐induced damage in subway tunnel structures remains a significant challenge due to the difficulty of full‐time and full‐area monitoring using traditional methods. This paper proposes a rapid approach for identifying and assessing the severity of cavity damage based on vibrations from moving trains. The methodology involves an advanced network training algorithm combining empirical mode decomposition (EMD), Hilbert transform, bidirectional long short−term memory (Bi‐LSTM), and self‐attention mechanisms. This novel combination enables the extraction of meaningful features from complex vibration data, facilitating precise damage detection. To validate the proposed method, a series of model tests with varying cavity sizes and locations was conducted. The three‐axis acceleration responses of a model train, recorded using a SmartRock sensor, were analyzed to evaluate the performance of the approach. Comparative analysis demonstrates the efficiency of the method, with the network achieving over 95% accuracy and a kappa coefficient above 0.95 for all classifiers in the validation set. The results confirm the potential of the proposed algorithm for rapid and accurate detection of cavity location and damage severity.

Recently, convolutional neural networks (CNNs) and hybrid networks, which integrate CNN with Transformer, have been widely employed in structuring crack detection, effectively addressing the challenges of high‐precision crack identification in controlled scenes. However, scene generalization remains a significant challenge for existing networks, especially under limited dataset conditions. With the rapid development of foundation models (like ChatGPT), achieving scene generalization has become feasible. In this paper, by taking tunnel crack detection as the background, the CraSAM network is proposed, which incorporates a foundation model‐based encoder and a prompt transfer learning module. Based on six datasets including tunnel, bridge, building, and pavement, the CraSAM is compared with 15 state‐of‐the‐art models, including Unet, DeepLabv3+, SSSeg, and TransUNet. It exhibits superior generalization capability both on few‐sample learned and unlearned conditions. This work will benefit to investigate of new ways for the utilization of the visual foundation model in various professional fields.