In recent years, video camera-based structural health monitoring has emerged as a viable method. With advanced cameras and software systems, achieving accurate results has become easier, and the accuracy rates of the obtained results have increased. In this study, a monitoring system composed of industrial cameras was developed. This system, consisting of four cameras, captured images from the minaret of Fatih Büyük Mosque, and displacement values were obtained through analysis. Using these displacement values, the structure’s frequencies and mode shapes were determined using the enhanced frequency domain decomposition method. A digital twin of the structure was created using the finite-element method, and frequency values were validated through modal analysis. Spectrum analysis identified the regions of stress accumulation. Finally, an earthquake simulation was conducted using the material properties obtained from optimization. The innovative aspect of this study lies in the use of a multiple camera system to focus on different points of the minaret, functioning like a sensor to perform nondestructive measurements. This approach enables detailed and precise displacement measurements without physically touching the structure, thus preserving its integrity. This study contributes to the literature on video-based systems used to obtain the frequencies and mode shapes of structures, with a particular focus on the structural health monitoring of historical masonry minarets. By doing so, it also makes a significant contribution to the preservation of historical structures.

Rapid visual screening (RVS) is a way to assess unsafe structures and reduce urban earthquake vulnerability; which is labor-intensive, time-consuming, and costly through conventional methods. Given that screening relies on visual clues, this study has utilized deep-learning convolutional neural networks including Residual Network (ResNet), Visual Geometry Group (VGG), Inception, and Densely Connected Convolutional Networks (DenseNet) to extract and classify necessary visual features to perform, enhance, and provide a novel standard computer vision–based method, complete with image databases and a computer program, to automate this procedure. This approach addresses the limitations of previous research in direct screening applications. The algorithm achieved 61%, 64%, 80%, 85%, and 96% accuracy in extracting various features, which, due to task independence, do not compromise its overall robustness. Furthermore, wielding hierarchical classification resulted in a noteworthy 18% accuracy improvement in building type classification. The proposed program produces final scores in 53 s, making it 17 times faster than conventional methods. Combined with its compatibility with portable computers, this will significantly enhance the speed and efficiency of onsite procedures, drastically reducing labor costs and making large-scale projects more feasible. Lastly, the severely damaged buildings classifier is exploitable for postearthquake fast screening and identification.

City-scale facility management plays a critical role in public health. This research aims to analyze city-scale infrastructure through a cost-effective approach by utilizing light detection and ranging (LiDAR) point cloud acquisition, dimensionality reduction processing, and advanced machine learning segmentation methods. A case study in a southwestern city in the United States examined bike lane infrastructure along 38.6 km using a backpack-mounted light detection and ranging LiDAR scanner on an electric scooter. The collected point cloud data were projected into two dimensions (2D) along the z-axis (up/down) and partitioned into crops in image format. A weakly supervised learning approach was employed for segmentation. The neural network segmented moving car trajectories, bike lane marking, and road curb segmentation. The postprocessed outputs evaluated city-scale bike infrastructure by measuring (1) lane widths; and (2) the closest distance between moving cars and the lanes, providing insights into the current state of road safety for cyclists. The results indicated that 85% of the roads examined had unsafe conditions due to reduced lane widths or shorter lateral distances, thus falling short of design guidelines. This study demonstrates a cost-effective and scalable method for city-scale data collection and analysis to inform infrastructure planning and cyclist safety assessments.

Accurate detection and quantification of damage to railway fasteners are crucial for ensuring railway safety. The spatial damage defects caused by the complex shape of fasteners and the problem of data imbalance in actual scenarios are significant challenges faced by deep learning models. This study innovatively proposes a railway-fastener point cloud analysis method based on deep learning as follows: (1) use four cameras to capture three-dimensional point cloud data and construct a virtual negative sample supplementary data set, (2) develop Rail-Swin3D models for precise segmentation of fastener components, and (3) introduce quantitative indicators to objectively evaluate the damage situation. A data set containing 120 real and virtual damaged fasteners was ultimately constructed, achieving up to 99.35% mean intersection over union (mIoU) in point cloud segmentation tasks. This study not only improves the efficiency of railway safety detection, but also opens new paths for the application of point cloud data in the field of railway maintenance, with profound theoretical and practical value.

Fog imposes adverse effect on driving safety. Traditional visibility measurement methods are expensive and limited to a short distance along the roadway. This study aims to identify visibility levels from foggy road images with deep learning methods. To address the shortage of foggy road image data set, a novel method is proposed to generate synthetic fog images based on point cloud and red, blue, & green (RGB) images. A synthetic foggy roadway image data set, kitti-foggy, containing 10,034 images was created with data from the kitti data set. Performance of the proposed method was compared with the traditional stereo-based method. Three typical image classification convolutional neural networks, including ResNet34, ResNet101, and Inception V4, were used to train the data set, and several evaluation matrices were used to evaluate their performances. The proposed method outputs more natural and authentic fog images. ResNet34 demonstrated the best performance among three algorithms with an overall accuracy of about 93%. Real data from a driving recorder and drones was used to verify the capability of ResNet34 to detect real fog. Findings of this study assist in the field of autonomous driving as well as intelligent transportation.

This study introduces a transformative artificial intelligence of things (AIoT) framework that advances bridge maintenance by incorporating advanced inspection techniques. A central innovation is the Pilgrimage Walk Optimization (PWO)-Lite algorithm, which fine-tunes the hyperparameters of the You Only Look Once (YOLO)v7-tiny deep learning model. This model, integrated with the Deep Simple Online and Realtime Tracking (DeepSORT) algorithm, enables real-time detection and significantly enhances the system’s ability to detect deteriorations in concrete beneath bridge decks swiftly and accurately. The PWO-Lite algorithm draws inspiration from the traditional Matsu pilgrimage, an important Taiwanese folk religious event. It reflects this influence in its search behavior, miming devotees’ gathering and movement patterns. This unique approach to algorithmic design incorporates cultural customs into computational strategies. An embedded system has been configured to efficiently process visual data from unmanned aerial vehicles (UAVs), providing actionable insights directly at the inspection site. This configuration reduces the reliance on heavy computational equipment and complex setups, streamlining bridge inspections and minimizing dependence on extensive infrastructure. The practical integration of this technology into UAVs allows engineers and field professionals to obtain precise, real-time data, enhancing maintenance planning and resource management. The broader implications of this research include the potential to significantly improve standard practices in infrastructure maintenance, offering a scalable solution that could revolutionize the field. This study bridges the gap between traditional AI applications and civil engineering. It introduces a culturally inspired optimization technique to structural health monitoring, benefiting both theoretical and practical aspects of infrastructure maintenance.

Modular construction is increasingly recognized for its efficiency in production and assembly. Modularization, the design process that segments floor plans into discrete modular units to create a modular key plan, has been identified as being crucial for creating an economical modular construction plan. However, solving the modularization problem is often daunting, requiring extensive trial-and-error to navigate the cast array of possible configurations. To address this challenge, this study introduces ModulePacking, a top-down generative design approach aimed specifically at the modularization problem. A hierarchical design methodology is proposed, encompassing a twofold process: partition and merging, where the optimized modular key plan is merged from the partition result. A genetic algorithm is utilized to accelerate the optimization process. In the case study, ModulePacking can generate modular key plans that rival those created by human designers as well as satisfactory results for large-scale buildings where human designers might struggle. ModulePacking showcases considerable potential in assisting designers and manufacturers in optimizing the modular construction plan, ultimately contributing to the advancement of modular construction.

With the development of information technology, building information modeling (BIM) is becoming widely used in the construction industry. It is very important to ensure the quality and compliance of BIM models. To solve the problem of time-consuming and laborious traditional checking methods and the limitation that existing software hard coding cannot adapt to rule changes, this paper proposes a BIM model quality compliance checking method based on knowledge graph that can quickly check BIM models through automated checking and ensure the modeling quality of BIM models. Firstly, knowledge extraction for BIM modeling standards and delivery standards is carried out based on knowledge graph technology to structurally represent text. Then, Revit secondary development technology and the C# programming language are adopted to develop a BIM model quality compliance checking system based on the knowledge graph analysis results and the structured query language (SQL) built-in rule knowledge base. Based on the knowledge base, the BIM model is automatically checked for quality compliance, and the naming, material, parameters, and rationality of each professional component are checked. Next, it is checked whether it conforms to the standard. Finally, the model modification plan is provided by checking the problem components and noncompliance details. The case application results show that the BIM model quality compliance checking system developed based on knowledge graph technology can identify defects or errors that do not meet the standards in the model, verify the feasibility and practicability of the system, and provide a new method for BIM designers to modify and improve the model.

The thorough investigation into the evolution of concrete performance under sulfate attack environments holds significant importance for engineering applications in specific conditions. In this paper, a prediction model for the two evaluation indexes of sulfate attack resistance of concrete (SARC), namely compressive strength corrosion resistance coefficient and mass loss rate, is established based on four machine-learning algorithms: Support Vector Regression, Random Forest Regression, Gradient Boosting, and Extreme Gradient Boosting (XGB). A comparison of the various performances showed that the model based on the XGB algorithm had the strongest generalization ability and offered the best prediction of SARC (K test set R2=0.963, MLR test set R2=0.903). Feature importance and partial correlation analyses were performed for the two XGB models separately, and a graphical user interface was designed based on the two predictive models. The results reveal that the number of cycles, water-binder ratio, and cement content significantly influence the SARC. Moderately increasing cement, fly ash, and coarse aggregate content can enhance the SARC. Increasing the number of cycles, drying time, water-binder ratio, sand, and solution concentration will reduce the SARC. Therefore, measures such as moderately increasing the amount of cement, reducing the water-binder ratio, and increasing the fly ash content can be increased to improve the SARC, but overuse has no significant effect.

In the dynamic construction environment, workers’ safety heavily depends on their ability to effectively perceive and react to hazards. Accordingly, studies have assessed the status of workers’ risk perception using advanced technologies. However, these studies have mainly focused on whether risks are perceived rather than how they are perceived. Recognizing the need for effective safety interventions that address risk-perception failures, it becomes crucial to not only classify workers’ risk-perception states but also to delve into the underlying processes of their risk perception. To address this research gap, this study examines the critical aspect of risk perception in construction safety by employing functional near-infrared spectroscopy (fNIRS) and 360° panoramas from actual construction sites to assess workers’ cognitive processes during hazard identification. Classifiers were developed using the AutoML method, and 15 advanced machine learning algorithms were compared to identify the highest-performing model. This model would then be utilized to understand the risk-perception process by incorporating the feature-importance technique. The results indicate that CatBoost emerged as the most effective classifier, achieving an accuracy rate of 90.3%. Additionally, the results identify significant brain activations in four anatomical locations: the prefrontal cortex, frontal eye fields, primary motor cortex, and primary auditory cortex. Notably, there is a significant correlation between these areas, emphasizing the importance of both visual and auditory cue perception in shaping workers’ situational awareness. This research highlights the potential of neuroimaging fNIRS in improving construction safety, and the importance of auditory perception in hazard identification, offering insights that could enhance the effectiveness of safety training programs.