Jon Olaizola, Unai Izagirre, Oscar Serradilla, Ekhi Zugasti, Mikel Mendicute, Jose I. Aizpurua
Ensuring the safe, reliable, and cost-efficient operation of transportation systems such as elevators is critical for the maintenance of civil infrastructures. The ability to monitor the health state and classify different operational states (elevator moving up/down, stopped, doors opening/closing) may lead to the development of intelligent solutions, such as diagnostics and predictive maintenance. Accordingly, downtime and maintenance costs can be significantly reduced with an accurate monitoring of the operation parameters and dynamics. In this context, this paper presents a novel approach for the operational state classification of elevator systems based on a one-dimensional convolutional neural network, using exclusively a single axis (Z) of an accelerometer signal. The proposed model utilizes a single accelerometer and addresses the challenge of distinguishing overlapping signal patterns, such as those produced by vertical displacement and door movements. The approach includes an interpretability stage, which demonstrates the data processing involved in extracting features from the underlying physical phenomena captured in the acceleration signal. Obtained results have been validated with an on-site captured dataset which contains 250 elevator journeys and compared with three other classification methods that have been conventionally used: generalized likelihood ratio test (GLRT), barometer-assisted GLRT, and three conventional machine learning modelss. It has been shown that the proposed approach is very accurate, with 96% of the average F1 score and, importantly, includes the analytic relation of the classification model features.
{"title":"An interpretable operational state classification framework for elevators through convolutional neural networks","authors":"Jon Olaizola, Unai Izagirre, Oscar Serradilla, Ekhi Zugasti, Mikel Mendicute, Jose I. Aizpurua","doi":"10.1111/mice.13479","DOIUrl":"https://doi.org/10.1111/mice.13479","url":null,"abstract":"Ensuring the safe, reliable, and cost-efficient operation of transportation systems such as elevators is critical for the maintenance of civil infrastructures. The ability to monitor the health state and classify different operational states (elevator moving up/down, stopped, doors opening/closing) may lead to the development of intelligent solutions, such as diagnostics and predictive maintenance. Accordingly, downtime and maintenance costs can be significantly reduced with an accurate monitoring of the operation parameters and dynamics. In this context, this paper presents a novel approach for the operational state classification of elevator systems based on a one-dimensional convolutional neural network, using exclusively a single axis (Z) of an accelerometer signal. The proposed model utilizes a single accelerometer and addresses the challenge of distinguishing overlapping signal patterns, such as those produced by vertical displacement and door movements. The approach includes an interpretability stage, which demonstrates the data processing involved in extracting features from the underlying physical phenomena captured in the acceleration signal. Obtained results have been validated with an on-site captured dataset which contains 250 elevator journeys and compared with three other classification methods that have been conventionally used: generalized likelihood ratio test (GLRT), barometer-assisted GLRT, and three conventional machine learning modelss. It has been shown that the proposed approach is very accurate, with 96% of the average F1 score and, importantly, includes the analytic relation of the classification model features.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"2 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143823136","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Zongyu Zhang, Junjie Huang, Qian Su, Shijie Liu, Naeem Mangi, Qi Zhang, Allen A. Zhang, Yao Liu, Shengyang Wang
The stability of embankment slopes for heavy-haul railway foundations is essential for safe railway operations. Railway embankment slope stability datasets often rely on engineering judgment for analysis. The labor- and resource-intensive processes of data preparation result in small dataset sizes. Machine learning analysis of small-sample potential features is a key low-cost approach for slope prediction. Due to the limited availability of slope failure data, a specialized framework is required for predictive modeling. To address this challenge, the focus is placed on data augmentation and interpretability analysis. A generative adversarial model is constructed using a graph convolutional network-based generator and a discriminator based on Gated Recurrent Unit, accompanied by a quality control method for the generated samples based on maximum mean discrepancy and one-class Support Vector Machine. This approach is designed to more effectively capture the temporal and spatial features of small samples. Three ensemble learning models, namely, XGBoost, random forest, and AdaBoost, are trained with augmented data, and model interpretation is conducted using Shapley Additive exPlanations to identify key factors affecting stability and potential stability improvement strategies. Results indicate that the proposed generative adversarial model surpasses traditional models in generating adequate data; the three enhanced data-trained machine learning models in this study achieved at least a 12% improvement in predictive accuracy, compared to their original small-sample-trained counterparts; The proposed data augmentation method outperformed variational autoencoder and diffusion models in generating high-quality synthetic data. Additionally, the interpretability framework effectively identified primary factors influencing slope stability. These findings provide a robust framework for interpretability-driven assessments of heavy-haul railway slopes with limited sample data.
{"title":"High embankment slope stability prediction using data augmentation and explainable ensemble learning","authors":"Zongyu Zhang, Junjie Huang, Qian Su, Shijie Liu, Naeem Mangi, Qi Zhang, Allen A. Zhang, Yao Liu, Shengyang Wang","doi":"10.1111/mice.13478","DOIUrl":"https://doi.org/10.1111/mice.13478","url":null,"abstract":"The stability of embankment slopes for heavy-haul railway foundations is essential for safe railway operations. Railway embankment slope stability datasets often rely on engineering judgment for analysis. The labor- and resource-intensive processes of data preparation result in small dataset sizes. Machine learning analysis of small-sample potential features is a key low-cost approach for slope prediction. Due to the limited availability of slope failure data, a specialized framework is required for predictive modeling. To address this challenge, the focus is placed on data augmentation and interpretability analysis. A generative adversarial model is constructed using a graph convolutional network-based generator and a discriminator based on Gated Recurrent Unit, accompanied by a quality control method for the generated samples based on maximum mean discrepancy and one-class Support Vector Machine. This approach is designed to more effectively capture the temporal and spatial features of small samples. Three ensemble learning models, namely, XGBoost, random forest, and AdaBoost, are trained with augmented data, and model interpretation is conducted using Shapley Additive exPlanations to identify key factors affecting stability and potential stability improvement strategies. Results indicate that the proposed generative adversarial model surpasses traditional models in generating adequate data; the three enhanced data-trained machine learning models in this study achieved at least a 12% improvement in predictive accuracy, compared to their original small-sample-trained counterparts; The proposed data augmentation method outperformed variational autoencoder and diffusion models in generating high-quality synthetic data. Additionally, the interpretability framework effectively identified primary factors influencing slope stability. These findings provide a robust framework for interpretability-driven assessments of heavy-haul railway slopes with limited sample data.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"183 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143823137","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Climate change exacerbates natural disasters, demanding rapid damage and risk assessment. However, expert-reliant analyses delay responses despite drone-aided data collection. This study develops and compares multimodal AI approaches using advanced large language models (LLMs) for expert-level landslide image analysis. We tackle landslide-specific challenges: capturing nuanced geotechnical reasoning beyond data digitization (specific to geological features and risk assessment), developing specialized transfer learning and data augmentation to mitigate data scarcity and geological diversity in landslide imagery, and establishing tailored evaluation metrics including geological accuracy, risk validity, and decision utility for landslide analysis. Evaluating a visual question answering-large language model (VQA-LLM) hybrid (sequential visual processing) and a multimodal large language model (MLLM, simultaneous vision/text processing) shows that MLLM excels in disaster identification, while the VQA-LLM hybrid demonstrates superior performance in risk assessment, thereby informing optimal AI design choices. Our methodology, structuring 30+ years of expert commentary for AI training and employing a comprehensive evaluation framework including standard text metrics, LLM-based semantic analysis, and expert domain assessment, highlights the potential of hybrid systems and addresses knowledge transfer in data-sparse domains.
{"title":"Multimodal artificial intelligence approaches using large language models for expert-level landslide image analysis","authors":"Kittitouch Areerob, Van-Quang Nguyen, Xianfeng Li, Shogo Inadomi, Toru Shimada, Hiroyuki Kanasaki, Zhijie Wang, Masanori Suganuma, Keiji Nagatani, Pang-jo Chun, Takayuki Okatani","doi":"10.1111/mice.13482","DOIUrl":"https://doi.org/10.1111/mice.13482","url":null,"abstract":"Climate change exacerbates natural disasters, demanding rapid damage and risk assessment. However, expert-reliant analyses delay responses despite drone-aided data collection. This study develops and compares multimodal AI approaches using advanced large language models (LLMs) for expert-level landslide image analysis. We tackle landslide-specific challenges: capturing nuanced geotechnical reasoning beyond data digitization (specific to geological features and risk assessment), developing specialized transfer learning and data augmentation to mitigate data scarcity and geological diversity in landslide imagery, and establishing tailored evaluation metrics including geological accuracy, risk validity, and decision utility for landslide analysis. Evaluating a visual question answering-large language model (VQA-LLM) hybrid (sequential visual processing) and a multimodal large language model (MLLM, simultaneous vision/text processing) shows that MLLM excels in disaster identification, while the VQA-LLM hybrid demonstrates superior performance in risk assessment, thereby informing optimal AI design choices. Our methodology, structuring 30+ years of expert commentary for AI training and employing a comprehensive evaluation framework including standard text metrics, LLM-based semantic analysis, and expert domain assessment, highlights the potential of hybrid systems and addresses knowledge transfer in data-sparse domains.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"12 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143823140","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
High-speed railway ballastless track structure crack detection usually has a high demand for the efficiency of crack detection technology. To overcome the limitation that current crack quantification methods usually require multiple steps, this paper proposes an efficient quantification method for track structure cracks using deep learning. This method applies the deep neural network (DNN) to the direct prediction of crack severity index values by modifying DNNs used for image classification. This method adopts a deep learning-based multi-step crack quantification method to calculate crack severity index values, establishes a dataset for predicting track structure interlayer crack severity index values using crack width mean values as labels, establishes a dataset for predicting track structure complex crack severity index values using crack width mean values and crack area values as labels, and utilizes the established datasets to train the modified DNNs. This method crops the track structure panorama in spatial order to obtain images, which not only facilitates DNN prediction but also enables the acquisition of more information such as crack distribution. Under the condition of using the data enhancement method, the mean absolute errors (MAEs) of the prediction results of the trained DNNs under the corresponding testing sets are 0.0191 and 0.0183, and the prediction results are in good agreement with the reference values. The image processing rates of the trained DNNs under the corresponding testing sets are all close to 75 images per second (resolution 512 × 512), which are 8.57 and 13.93 times as computationally efficient as the adopted deep learning-based multi-step crack quantification method.
{"title":"Efficient quantifying track structure cracks using deep learning","authors":"Hongshuo Sun, Li Song, Zhiwu Yu","doi":"10.1111/mice.13477","DOIUrl":"https://doi.org/10.1111/mice.13477","url":null,"abstract":"High-speed railway ballastless track structure crack detection usually has a high demand for the efficiency of crack detection technology. To overcome the limitation that current crack quantification methods usually require multiple steps, this paper proposes an efficient quantification method for track structure cracks using deep learning. This method applies the deep neural network (DNN) to the direct prediction of crack severity index values by modifying DNNs used for image classification. This method adopts a deep learning-based multi-step crack quantification method to calculate crack severity index values, establishes a dataset for predicting track structure interlayer crack severity index values using crack width mean values as labels, establishes a dataset for predicting track structure complex crack severity index values using crack width mean values and crack area values as labels, and utilizes the established datasets to train the modified DNNs. This method crops the track structure panorama in spatial order to obtain images, which not only facilitates DNN prediction but also enables the acquisition of more information such as crack distribution. Under the condition of using the data enhancement method, the mean absolute errors (MAEs) of the prediction results of the trained DNNs under the corresponding testing sets are 0.0191 and 0.0183, and the prediction results are in good agreement with the reference values. The image processing rates of the trained DNNs under the corresponding testing sets are all close to 75 images per second (resolution 512 × 512), which are 8.57 and 13.93 times as computationally efficient as the adopted deep learning-based multi-step crack quantification method.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"31 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2025-04-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143814185","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper proposes a key origin–destination (OD) pairs perception reasoning (KODPR) approach for route guidance (RG) in urban traffic networks with numerous OD pairs. First, to reduce a real‐world RG problem's complexity with large OD sizes, a long‐term perception module is developed to identify a few critical OD pairs, making real‐world application feasible. Second, the issue of multi‐OD cooperation and system resource allocation is addressed through the cooperative perception reasoning method that performs a sequential action update mechanism among agents. Additionally, a balanced reward function is designed in the Markov decision process framework for optimizing dynamic RG strategies. Experimental results using a real‐world road network in Hangzhou, China, within a simulation of urban mobility‐based simulation platform, demonstrate the superior performance of the proposed approach. The KODPR achieves optimization results close to dynamic user equilibrium by adjusting only 30% of the OD pairs in the network, significantly outperforming comparison methods. Its ability to coordinate extensive OD pairs in densely populated urban environments presents a promising solution for urban traffic RG.
{"title":"Key origin–destination pairs perception reasoning approach","authors":"Zheyuan Jiang, Ziyi Shi, Zheng Zhu, Xiqun (Michael) Chen","doi":"10.1111/mice.13476","DOIUrl":"https://doi.org/10.1111/mice.13476","url":null,"abstract":"This paper proposes a key origin–destination (OD) pairs perception reasoning (KODPR) approach for route guidance (RG) in urban traffic networks with numerous OD pairs. First, to reduce a real‐world RG problem's complexity with large OD sizes, a long‐term perception module is developed to identify a few critical OD pairs, making real‐world application feasible. Second, the issue of multi‐OD cooperation and system resource allocation is addressed through the cooperative perception reasoning method that performs a sequential action update mechanism among agents. Additionally, a balanced reward function is designed in the Markov decision process framework for optimizing dynamic RG strategies. Experimental results using a real‐world road network in Hangzhou, China, within a simulation of urban mobility‐based simulation platform, demonstrate the superior performance of the proposed approach. The KODPR achieves optimization results close to dynamic user equilibrium by adjusting only 30% of the OD pairs in the network, significantly outperforming comparison methods. Its ability to coordinate extensive OD pairs in densely populated urban environments presents a promising solution for urban traffic RG.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"1 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2025-04-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143813601","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This study presents a scalable multi‐camera system (S‐MCS) for high‐precision displacement measurement and deformation monitoring of long‐span arch bridges during construction. Traditional methods such as robotic total stations (RTS) and single‐camera systems face limitations in dynamic scalability, synchronous multi‐point monitoring, and robustness against environmental disturbances. To address these challenges, the proposed S‐MCS integrates dynamically expandable measuring cameras and dual correcting cameras to compensate for platform ego‐motion. A self‐calibration algorithm and spatiotemporal reference alignment framework are developed to ensure measurement consistency across evolving construction phases. The system was deployed on a 600‐m‐span arch bridge, achieving sub‐millimeter accuracy (root mean square error ≤ 1.09 mm) validated against RTS data. Key innovations include real‐time platform motion compensation, adaptive coverage expansion, and high‐frequency sampling for capturing transient structural responses. Comparative analyses under construction loads, thermal variations, and extreme crosswinds demonstrated the system's superiority in tracking multi‐point displacements, resolving dynamic behaviors and supporting safety assessments. The S‐MCS provides a robust solution for automated, large‐scale structural health monitoring, with potential applications in diverse infrastructure projects requiring adaptive, high‐resolution deformation tracking.
{"title":"A displacement measurement methodology for deformation monitoring of long‐span arch bridges during construction based on scalable multi‐camera system","authors":"Yihe Yin, Xiaolin Liu, Biao Hu, Wenjun Chen, Xiao Guo, Danyang Ma, Xiaohua Ding, Linhai Han, Qifeng Yu","doi":"10.1111/mice.13475","DOIUrl":"https://doi.org/10.1111/mice.13475","url":null,"abstract":"This study presents a scalable multi‐camera system (S‐MCS) for high‐precision displacement measurement and deformation monitoring of long‐span arch bridges during construction. Traditional methods such as robotic total stations (RTS) and single‐camera systems face limitations in dynamic scalability, synchronous multi‐point monitoring, and robustness against environmental disturbances. To address these challenges, the proposed S‐MCS integrates dynamically expandable measuring cameras and dual correcting cameras to compensate for platform ego‐motion. A self‐calibration algorithm and spatiotemporal reference alignment framework are developed to ensure measurement consistency across evolving construction phases. The system was deployed on a 600‐m‐span arch bridge, achieving sub‐millimeter accuracy (root mean square error ≤ 1.09 mm) validated against RTS data. Key innovations include real‐time platform motion compensation, adaptive coverage expansion, and high‐frequency sampling for capturing transient structural responses. Comparative analyses under construction loads, thermal variations, and extreme crosswinds demonstrated the system's superiority in tracking multi‐point displacements, resolving dynamic behaviors and supporting safety assessments. The S‐MCS provides a robust solution for automated, large‐scale structural health monitoring, with potential applications in diverse infrastructure projects requiring adaptive, high‐resolution deformation tracking.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"34 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143782504","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The cover image is based on the article A machine vision-based intelligent segmentation method for dam underwater cracks using swarm optimization algorithm and deep learning by Yantao Zhu et al., https://doi.org/10.1111/mice.13343.
{"title":"Cover Image, Volume 40, Issue 10","authors":"","doi":"10.1111/mice.13473","DOIUrl":"https://doi.org/10.1111/mice.13473","url":null,"abstract":"<p><b>The cover image</b> is based on the article <i>A machine vision-based intelligent segmentation method for dam underwater cracks using swarm optimization algorithm and deep learning</i> by Yantao Zhu et al., https://doi.org/10.1111/mice.13343.\u0000\u0000 <figure>\u0000 <div><picture>\u0000 <source></source></picture><p></p>\u0000 </div>\u0000 </figure></p>","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"40 10","pages":""},"PeriodicalIF":8.5,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1111/mice.13473","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143778284","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The cover image is based on the article Short-term Prediction of Railway Track Degradation Using Ensemble Deep Learning by Yong Zhuang et al., https://doi.org/10.1111/mice.13462.
{"title":"Cover Image, Volume 40, Issue 10","authors":"","doi":"10.1111/mice.13474","DOIUrl":"https://doi.org/10.1111/mice.13474","url":null,"abstract":"<p><b>The cover image</b> is based on the article <i>Short-term Prediction of Railway Track Degradation Using Ensemble Deep Learning</i> by Yong Zhuang et al., https://doi.org/10.1111/mice.13462.\u0000\u0000 <figure>\u0000 <div><picture>\u0000 <source></source></picture><p></p>\u0000 </div>\u0000 </figure></p>","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"40 10","pages":""},"PeriodicalIF":8.5,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1111/mice.13474","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143778280","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Roof inspections are crucial but perilous, necessitating safer and more cost-effective solutions. While robots offer promising solutions to reduce fall risks, robotic vision systems face efficiency limitations due to computational constraints and scarce specialized data. This study presents real-time roof defect segmentation network (RRD-SegNet), a deep learning framework optimized for mobile robotic platforms. The architecture features a mobile-efficient backbone network for lightweight processing, a defect-specific feature extraction module for improved accuracy, and a regressive detection and classification head for precise defect localization. Trained on the multi-type roof defect segmentation dataset of 1350 annotated images across six defect categories, RRD-SegNet integrates with a roof damage identification module for real-time tracking. The system surpasses state-of-the-art models with 85.2% precision and 76.8% recall while requiring minimal computational resources. Field testing confirms its effectiveness with F1-scores of 0.720–0.945 across defect types at processing speeds of 1.62 ms/frame. This work advances automated inspection in civil engineering by enabling efficient, safe, and accurate roof assessments via mobile robotic platforms.
{"title":"A computational method for real-time roof defect segmentation in robotic inspection","authors":"Xiayu Zhao, Houtan Jebelli","doi":"10.1111/mice.13471","DOIUrl":"https://doi.org/10.1111/mice.13471","url":null,"abstract":"Roof inspections are crucial but perilous, necessitating safer and more cost-effective solutions. While robots offer promising solutions to reduce fall risks, robotic vision systems face efficiency limitations due to computational constraints and scarce specialized data. This study presents real-time roof defect segmentation network (RRD-SegNet), a deep learning framework optimized for mobile robotic platforms. The architecture features a mobile-efficient backbone network for lightweight processing, a defect-specific feature extraction module for improved accuracy, and a regressive detection and classification head for precise defect localization. Trained on the multi-type roof defect segmentation dataset of 1350 annotated images across six defect categories, RRD-SegNet integrates with a roof damage identification module for real-time tracking. The system surpasses state-of-the-art models with 85.2% precision and 76.8% recall while requiring minimal computational resources. Field testing confirms its effectiveness with F1-scores of 0.720–0.945 across defect types at processing speeds of 1.62 ms/frame. This work advances automated inspection in civil engineering by enabling efficient, safe, and accurate roof assessments via mobile robotic platforms.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"73 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2025-04-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143776044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To overcome the limitations of fragility analysis in the assessment of partition walls, specifically data shortage, general uncertainties, and subjective criteria, this study proposes a probabilistic method to evaluate seismic damage of partition walls. A proposed multi-spring numerical model balances the damage representation and computational efficiency in simulations, thus avoiding extensive experimental testing. By accounting for parameter uncertainties in individual partition walls, the uncertainties introduced by the fragility group are avoided, and the description of the seismic damage is probabilistic, enhancing the reliability of the assessment results. Using damaged areas as the assessment criterion alleviates epistemic uncertainty exacerbated by subjective judgments on repair actions. Furthermore, it eliminates the assumption of a log-normal distribution for damage in fragility analysis, improving the calculations of damage probabilities and expected repair costs. The results are anticipated to be valuable for assessing the seismic risk and repair costs of partition walls.
{"title":"Probabilistic seismic damage assessment for partition walls based on a multi-spring numerical model incorporating uncertainties","authors":"Jiantao Huang, Masahiro Kurata","doi":"10.1111/mice.13472","DOIUrl":"https://doi.org/10.1111/mice.13472","url":null,"abstract":"To overcome the limitations of fragility analysis in the assessment of partition walls, specifically data shortage, general uncertainties, and subjective criteria, this study proposes a probabilistic method to evaluate seismic damage of partition walls. A proposed multi-spring numerical model balances the damage representation and computational efficiency in simulations, thus avoiding extensive experimental testing. By accounting for parameter uncertainties in individual partition walls, the uncertainties introduced by the fragility group are avoided, and the description of the seismic damage is probabilistic, enhancing the reliability of the assessment results. Using damaged areas as the assessment criterion alleviates epistemic uncertainty exacerbated by subjective judgments on repair actions. Furthermore, it eliminates the assumption of a log-normal distribution for damage in fragility analysis, improving the calculations of damage probabilities and expected repair costs. The results are anticipated to be valuable for assessing the seismic risk and repair costs of partition walls.","PeriodicalId":156,"journal":{"name":"Computer-Aided Civil and Infrastructure Engineering","volume":"11 1","pages":""},"PeriodicalIF":11.775,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143713646","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}