Pub Date : 2024-10-26DOI: 10.1016/j.autcon.2024.105839
Andong Qiu, Zhouwang Yang
Distributed photovoltaic power systems, typically deployed in complex scenarios like irregular rooftops, present a challenging detailed cable routing problem (DCRP). This involves grouping solar modules and routing cables to connect each group, traditionally addressed through manual design. This paper presents a variable-depth large neighborhood search (VDLNS) algorithm to address the DCRP, which is modeled as a specialized cycle covering problem using arc-flow and partition formulations. A cycle-split heuristic, derived from DCRP’s connection to the traveling salesman problem, is introduced and combined with a series of destroy operators to construct the VDLNS algorithm. Numerical experiments conducted on both synthetic and real-world instances validated the algorithm’s efficacy, achieving an average total cost reduction of 12.87% on house rooftop instances compared to manual design. The results indicate that the method effectively streamlines photovoltaic system design by delivering cost-efficient cable routing schemes within a reasonable timeframe.
{"title":"Variable-depth large neighborhood search algorithm for cable routing in distributed photovoltaic systems","authors":"Andong Qiu, Zhouwang Yang","doi":"10.1016/j.autcon.2024.105839","DOIUrl":"10.1016/j.autcon.2024.105839","url":null,"abstract":"<div><div>Distributed photovoltaic power systems, typically deployed in complex scenarios like irregular rooftops, present a challenging detailed cable routing problem (DCRP). This involves grouping solar modules and routing cables to connect each group, traditionally addressed through manual design. This paper presents a variable-depth large neighborhood search (VDLNS) algorithm to address the DCRP, which is modeled as a specialized cycle covering problem using arc-flow and partition formulations. A cycle-split heuristic, derived from DCRP’s connection to the traveling salesman problem, is introduced and combined with a series of destroy operators to construct the VDLNS algorithm. Numerical experiments conducted on both synthetic and real-world instances validated the algorithm’s efficacy, achieving an average total cost reduction of 12.87% on house rooftop instances compared to manual design. The results indicate that the method effectively streamlines photovoltaic system design by delivering cost-efficient cable routing schemes within a reasonable timeframe.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105839"},"PeriodicalIF":9.6,"publicationDate":"2024-10-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142531375","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}
Pub Date : 2024-10-24DOI: 10.1016/j.autcon.2024.105843
Gyumin Lee , Ali Turab Asad , Khurram Shabbir , Sung-Han Sim , Junhwa Lee
Accelerated bridge construction (ABC) demands precise alignment of prefabricated members to prevent assembly failure. Conventional methods struggle to localize shear connectors from point cloud data (PCD) generated by structure-from-motion due to its sparsity. This paper introduces a robust method for shear connector localization using PCD generated by a neural radiance field and a three-step narrowing-down algorithm. The PCD exhibits densely populated points for small connectors, allowing the algorithm to pinpoint their locations accurately. The method successfully identified all 72 shear connectors in a mock-up prefabricated girder, with an average error of 10 mm, demonstrating its potential for assessing constructability in ABC projects. Future research may integrate deep learning-based segmentation techniques to enhance efficiency and adaptability in complex geometries and non-standard bridge designs.
{"title":"Robust localization of shear connectors in accelerated bridge construction with neural radiance field","authors":"Gyumin Lee , Ali Turab Asad , Khurram Shabbir , Sung-Han Sim , Junhwa Lee","doi":"10.1016/j.autcon.2024.105843","DOIUrl":"10.1016/j.autcon.2024.105843","url":null,"abstract":"<div><div>Accelerated bridge construction (ABC) demands precise alignment of prefabricated members to prevent assembly failure. Conventional methods struggle to localize shear connectors from point cloud data (PCD) generated by structure-from-motion due to its sparsity. This paper introduces a robust method for shear connector localization using PCD generated by a neural radiance field and a three-step narrowing-down algorithm. The PCD exhibits densely populated points for small connectors, allowing the algorithm to pinpoint their locations accurately. The method successfully identified all 72 shear connectors in a mock-up prefabricated girder, with an average error of 10 mm, demonstrating its potential for assessing constructability in ABC projects. Future research may integrate deep learning-based segmentation techniques to enhance efficiency and adaptability in complex geometries and non-standard bridge designs.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105843"},"PeriodicalIF":9.6,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142531374","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}
Pub Date : 2024-10-24DOI: 10.1016/j.autcon.2024.105844
SeyedeZahra Golazad , Abbas Mohammadi , Abbas Rashidi , Mohammad Ilbeigi
As the use of predictive models in construction rapidly increases, the need for preprocessing raw construction data has become more critical. This systematic review investigates data preprocessing techniques for machine learning (ML), deep learning (DL), and reinforcement learning (RL) models in the construction domain. Through a comprehensive analysis of 457 studies, the prevalence of six data types (i.e., tabular, image, video frame, time series, text, and point cloud) and their respective preprocessing methods are examined. Key findings reveal data transformation, cleaning, reduction, augmentation, and scaling as fundamental preprocessing categories, with applications varying across data types. The paper highlights knowledge gaps, including limited synthetic data adoption, lack of standardized annotation practices, absence of comprehensive preprocessing frameworks, and need for automated labeling. Furthermore, critical considerations regarding data privacy, security, sharing, and management practices are discussed. The review underscores the pivotal role of robust data preprocessing in enabling reliable predictive models.
{"title":"From raw to refined: Data preprocessing for construction machine learning (ML), deep learning (DL), and reinforcement learning (RL) models","authors":"SeyedeZahra Golazad , Abbas Mohammadi , Abbas Rashidi , Mohammad Ilbeigi","doi":"10.1016/j.autcon.2024.105844","DOIUrl":"10.1016/j.autcon.2024.105844","url":null,"abstract":"<div><div>As the use of predictive models in construction rapidly increases, the need for preprocessing raw construction data has become more critical. This systematic review investigates data preprocessing techniques for machine learning (ML), deep learning (DL), and reinforcement learning (RL) models in the construction domain. Through a comprehensive analysis of 457 studies, the prevalence of six data types (i.e., tabular, image, video frame, time series, text, and point cloud) and their respective preprocessing methods are examined. Key findings reveal data transformation, cleaning, reduction, augmentation, and scaling as fundamental preprocessing categories, with applications varying across data types. The paper highlights knowledge gaps, including limited synthetic data adoption, lack of standardized annotation practices, absence of comprehensive preprocessing frameworks, and need for automated labeling. Furthermore, critical considerations regarding data privacy, security, sharing, and management practices are discussed. The review underscores the pivotal role of robust data preprocessing in enabling reliable predictive models.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105844"},"PeriodicalIF":9.6,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534959","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}
Pub Date : 2024-10-24DOI: 10.1016/j.autcon.2024.105837
Qishen Ye, Yihai Fang, Nan Zheng
Road work zones pose significant safety risks to both vehicles passing by and the construction workers moving within the work zones. Over recent years, significant research efforts have been dedicated to work zone safety, particularly by leveraging emerging technologies. This paper aims to review the literature on performance evaluation of safety technologies designed to mitigate struck-by hazards. This review identified 57 relevant publications focusing on technology evaluation, which were critically reviewed using the Four-component Cyber-Physical System (CPS) hierarchy and the Adapted Layer of Protection Analysis (ALOPA) framework. The CPS hierarchy-based review unveiled the focused components under evaluation, the relationship among these components, the methodologies employed, and the key performance results. The extent and completeness of the evaluation methods were examined through the ALOPA framework. The findings of this research highlight emerging trends that explore the impact of human factors on accident avoidance outcomes in risk-free virtual environments and suggest several prospective considerations as per ALOPA that can guide future research towards performance-based evaluations and design optimisations.
{"title":"Performance evaluation of struck-by-accident alert systems for road work zone safety","authors":"Qishen Ye, Yihai Fang, Nan Zheng","doi":"10.1016/j.autcon.2024.105837","DOIUrl":"10.1016/j.autcon.2024.105837","url":null,"abstract":"<div><div>Road work zones pose significant safety risks to both vehicles passing by and the construction workers moving within the work zones. Over recent years, significant research efforts have been dedicated to work zone safety, particularly by leveraging emerging technologies. This paper aims to review the literature on performance evaluation of safety technologies designed to mitigate struck-by hazards. This review identified 57 relevant publications focusing on technology evaluation, which were critically reviewed using the Four-component Cyber-Physical System (CPS) hierarchy and the Adapted Layer of Protection Analysis (ALOPA) framework. The CPS hierarchy-based review unveiled the focused components under evaluation, the relationship among these components, the methodologies employed, and the key performance results. The extent and completeness of the evaluation methods were examined through the ALOPA framework. The findings of this research highlight emerging trends that explore the impact of human factors on accident avoidance outcomes in risk-free virtual environments and suggest several prospective considerations as per ALOPA that can guide future research towards performance-based evaluations and design optimisations.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105837"},"PeriodicalIF":9.6,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534960","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}
Pub Date : 2024-10-23DOI: 10.1016/j.autcon.2024.105834
M. Franciosi, M. Kasser, M. Viviani
Digital twins are evolving to oversee the entire construction life cycle, with a strong emphasis on sustainability across environmental, financial, regulatory, and administrative dimensions. This paper introduces a methodology for managing existing bridges through an adaptable digital twin. The aim of this research is to develop a framework for constructing digital twins that, by enabling structural analysis and “what-if” scenario simulations, supports more reliable maintenance decision-making. Such type of digital twin ensure safety, extend lifespan, and provide a precise database for managing end-of-life processes within a circular “cradle to cradle” framework. This methodology also addresses obsolescence issues related to software evolution and the longer lifespan of a bridge compared to its creator. A case study demonstrates the methodology's effectiveness, showing that digital twins can be flexible, cost-effective tools for managing all types of bridges, including small and existing ones.
{"title":"Digital twins in bridge engineering for streamlined maintenance and enhanced sustainability","authors":"M. Franciosi, M. Kasser, M. Viviani","doi":"10.1016/j.autcon.2024.105834","DOIUrl":"10.1016/j.autcon.2024.105834","url":null,"abstract":"<div><div>Digital twins are evolving to oversee the entire construction life cycle, with a strong emphasis on sustainability across environmental, financial, regulatory, and administrative dimensions. This paper introduces a methodology for managing existing bridges through an adaptable digital twin. The aim of this research is to develop a framework for constructing digital twins that, by enabling structural analysis and “what-if” scenario simulations, supports more reliable maintenance decision-making. Such type of digital twin ensure safety, extend lifespan, and provide a precise database for managing end-of-life processes within a circular “cradle to cradle” framework. This methodology also addresses obsolescence issues related to software evolution and the longer lifespan of a bridge compared to its creator. A case study demonstrates the methodology's effectiveness, showing that digital twins can be flexible, cost-effective tools for managing all types of bridges, including small and existing ones.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105834"},"PeriodicalIF":9.6,"publicationDate":"2024-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534957","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}
Pub Date : 2024-10-23DOI: 10.1016/j.autcon.2024.105845
Xin Zuo , Yu Sheng , Jifeng Shen , Yongwei Shan
CrackMamba, a Mamba-based model, is designed for efficient and accurate crack segmentation for monitoring the structural health of infrastructure. Traditional Convolutional Neural Network (CNN) models struggle with limited receptive fields, and while Vision Transformers (ViT) improve segmentation accuracy, they are computationally intensive. CrackMamba addresses these challenges by utilizing the VMambaV2 with pre-trained ImageNet-1 k weights as the encoder and a newly designed decoder for better performance. To handle the random and complex nature of crack development, a Snake Scan module is proposed to reshape crack feature sequences, enhancing feature extraction. Additionally, the three-branch Snake Conv VSS (SCVSS) block is proposed to target cracks more effectively. Experiments show that CrackMamba achieves state-of-the-art (SOTA) performance on the CrackSeg9k and SewerCrack datasets, and demonstrates competitive performance on the retinal vessel segmentation dataset CHASE_DB1, highlighting its generalization capability.
{"title":"Topology-aware mamba for crack segmentation in structures","authors":"Xin Zuo , Yu Sheng , Jifeng Shen , Yongwei Shan","doi":"10.1016/j.autcon.2024.105845","DOIUrl":"10.1016/j.autcon.2024.105845","url":null,"abstract":"<div><div>CrackMamba, a Mamba-based model, is designed for efficient and accurate crack segmentation for monitoring the structural health of infrastructure. Traditional Convolutional Neural Network (CNN) models struggle with limited receptive fields, and while Vision Transformers (ViT) improve segmentation accuracy, they are computationally intensive. CrackMamba addresses these challenges by utilizing the VMambaV2 with pre-trained ImageNet-1 k weights as the encoder and a newly designed decoder for better performance. To handle the random and complex nature of crack development, a Snake Scan module is proposed to reshape crack feature sequences, enhancing feature extraction. Additionally, the three-branch Snake Conv VSS (SCVSS) block is proposed to target cracks more effectively. Experiments show that CrackMamba achieves state-of-the-art (SOTA) performance on the CrackSeg9k and SewerCrack datasets, and demonstrates competitive performance on the retinal vessel segmentation dataset CHASE_DB1, highlighting its generalization capability.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105845"},"PeriodicalIF":9.6,"publicationDate":"2024-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534958","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}
Pub Date : 2024-10-23DOI: 10.1016/j.autcon.2024.105823
Sungmin Yoon , Jeyoon Lee , Jiteng Li , Peng Wang
A virtual model that mathematically represents operational behaviors is essential for implementing the concepts of digital twins (DTs) and building information modeling (BIM) to achieve intelligent, optimal building operations. However, current research lacks an approach to reliably construct virtual models. This paper introduces a concept named virtual in-situ modeling (VIM), designed to comprehensively represent building behaviors. The VIM framework is based on five key aspects: modeling environments, model types, modeling sources, modeling approaches, and model fusion techniques. VIM bridges BIM and DT, enabling virtual modeling during the operational phase and enhancing both BIM-based DT and DT-enhanced BIM. Case studies conducted using real building operations demonstrate the effectiveness of VIM, achieving a highly accuracy (RMSE of 0.24 °C). Additionally, the VIM-assisted fault detection and diagnosis (FDD) provided early detection and diagnostic estimation, outperforming FDD without the virtual model. This paper highlights the potential of VIM for advanced building operations and maintenance.
{"title":"Virtual in-situ modeling between digital twin and BIM for advanced building operations and maintenance","authors":"Sungmin Yoon , Jeyoon Lee , Jiteng Li , Peng Wang","doi":"10.1016/j.autcon.2024.105823","DOIUrl":"10.1016/j.autcon.2024.105823","url":null,"abstract":"<div><div>A virtual model that mathematically represents operational behaviors is essential for implementing the concepts of digital twins (DTs) and building information modeling (BIM) to achieve intelligent, optimal building operations. However, current research lacks an approach to reliably construct virtual models. This paper introduces a concept named virtual in-situ modeling (VIM), designed to comprehensively represent building behaviors. The VIM framework is based on five key aspects: modeling environments, model types, modeling sources, modeling approaches, and model fusion techniques. VIM bridges BIM and DT, enabling virtual modeling during the operational phase and enhancing both BIM-based DT and DT-enhanced BIM. Case studies conducted using real building operations demonstrate the effectiveness of VIM, achieving a highly accuracy (RMSE of 0.24 °C). Additionally, the VIM-assisted fault detection and diagnosis (FDD) provided early detection and diagnostic estimation, outperforming FDD without the virtual model. This paper highlights the potential of VIM for advanced building operations and maintenance.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105823"},"PeriodicalIF":9.6,"publicationDate":"2024-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534956","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}
Pub Date : 2024-10-22DOI: 10.1016/j.autcon.2024.105828
B.G. Pantoja-Rosero , A. Chassignet , A. Rezaie , M. Kozinski , R. Achanta , K. Beyer
Semantic segmentation with deep learning plays a crucial role in various fields, including civil engineering, particularly in tasks such as damage assessment and urban planning. This paper addresses the challenge of efficiently training deep learning models for semantic segmentation with a limited set of annotated data, thus reducing the burden of ground truth labeling. An active learning strategy is introduced, leveraging partial annotations informed by predictions and uncertainties from previously trained models. Unlike other active learning frameworks, this approach not only facilitates the annotation of highly uncertain image regions but also targets those with low uncertainty, which often lead to false positives and negatives. The results demonstrate that using partial annotations within an active learning framework significantly reduces manual annotation efforts and training time without compromising model performance. These findings have substantial implications for the efficiency and scalability of deep learning in civil engineering, paving the way for future research in active learning and semantic segmentation.
{"title":"Partial annotations in active learning for semantic segmentation","authors":"B.G. Pantoja-Rosero , A. Chassignet , A. Rezaie , M. Kozinski , R. Achanta , K. Beyer","doi":"10.1016/j.autcon.2024.105828","DOIUrl":"10.1016/j.autcon.2024.105828","url":null,"abstract":"<div><div>Semantic segmentation with deep learning plays a crucial role in various fields, including civil engineering, particularly in tasks such as damage assessment and urban planning. This paper addresses the challenge of efficiently training deep learning models for semantic segmentation with a limited set of annotated data, thus reducing the burden of ground truth labeling. An active learning strategy is introduced, leveraging partial annotations informed by predictions and uncertainties from previously trained models. Unlike other active learning frameworks, this approach not only facilitates the annotation of highly uncertain image regions but also targets those with low uncertainty, which often lead to false positives and negatives. The results demonstrate that using partial annotations within an active learning framework significantly reduces manual annotation efforts and training time without compromising model performance. These findings have substantial implications for the efficiency and scalability of deep learning in civil engineering, paving the way for future research in active learning and semantic segmentation.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105828"},"PeriodicalIF":9.6,"publicationDate":"2024-10-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534919","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}
Pub Date : 2024-10-22DOI: 10.1016/j.autcon.2024.105814
Zhengkai Zhao , Shu Zhang , Xinyu Hua , Xiuzhi Shi
Understanding how workers perceive risk is essential to construction safety management. Firstly, an event-related potential (ERP) experiment was conducted to investigate the relationship between risk, likelihood, and severity. Then, a linear model was developed to predict workers' risk perception based on ERP components and quantify the relative importance of severity to likelihood. Finally, an additive model was constructed to reflect the risk perception pattern. The results indicate: (1) Workers' emotional responses stem from the process of associating accident consequences in severity assessment, which is represented by the late positive potential (LPP) component. (2) Workers' risk perception relies more on severity compared with likelihood. (3) The additive model (risk = 0.203 * likelihood +0.758 * severity) better matches the risk perception patterns than the multiplicative model. The research results provide a new perspective for understanding workers' risk perception patterns and contributing to proactive safety management in the construction industry.
{"title":"Investigating construction workers' perception of risk, likelihood, and severity using electroencephalogram and machine learning","authors":"Zhengkai Zhao , Shu Zhang , Xinyu Hua , Xiuzhi Shi","doi":"10.1016/j.autcon.2024.105814","DOIUrl":"10.1016/j.autcon.2024.105814","url":null,"abstract":"<div><div>Understanding how workers perceive risk is essential to construction safety management. Firstly, an event-related potential (ERP) experiment was conducted to investigate the relationship between risk, likelihood, and severity. Then, a linear model was developed to predict workers' risk perception based on ERP components and quantify the relative importance of severity to likelihood. Finally, an additive model was constructed to reflect the risk perception pattern. The results indicate: (1) Workers' emotional responses stem from the process of associating accident consequences in severity assessment, which is represented by the late positive potential (LPP) component. (2) Workers' risk perception relies more on severity compared with likelihood. (3) The additive model (risk = 0.203 * likelihood +0.758 * severity) better matches the risk perception patterns than the multiplicative model. The research results provide a new perspective for understanding workers' risk perception patterns and contributing to proactive safety management in the construction industry.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105814"},"PeriodicalIF":9.6,"publicationDate":"2024-10-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534955","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}
Pub Date : 2024-10-21DOI: 10.1016/j.autcon.2024.105830
Shengli Li , Shiji Sun , Yang Liu , Wanshuai Qi , Nan Jiang , Can Cui , Pengfei Zheng
It is challenging to distinguish the defective areas using infrared thermography to automatically analyze external post-tensioned tendon duct grouting defects. To achieve efficient and stable automated detection, a lightweight real-time grouting defects detection method based on YOLO deep learning is proposed. Firstly, the Cutpaste data augmentation method was used to effectively alleviate the problem of overfitting. Then, the C3Ghost module was introduced into the neck network, and the number of channels in the network layers was adjusted to 50 % of those in the YOLOv5s model, reducing the number of parameters and computational resources. Finally, the SGD optimizer and GIOU loss function, as well as the Sim attention module, were used to improve detection accuracy. Based on instance analysis and comparison, this method achieves [email protected] of 96.9 % and detection speed of 66FPS. Compared with YOLOv5s, it reduces the number of parameters by 79 % and FLOPs by 77 %.
{"title":"Real-time lightweight YOLO model for grouting defect detection in external post-tensioned ducts via infrared thermography","authors":"Shengli Li , Shiji Sun , Yang Liu , Wanshuai Qi , Nan Jiang , Can Cui , Pengfei Zheng","doi":"10.1016/j.autcon.2024.105830","DOIUrl":"10.1016/j.autcon.2024.105830","url":null,"abstract":"<div><div>It is challenging to distinguish the defective areas using infrared thermography to automatically analyze external post-tensioned tendon duct grouting defects. To achieve efficient and stable automated detection, a lightweight real-time grouting defects detection method based on YOLO deep learning is proposed. Firstly, the Cutpaste data augmentation method was used to effectively alleviate the problem of overfitting. Then, the C3Ghost module was introduced into the neck network, and the number of channels in the network layers was adjusted to 50 % of those in the YOLOv5s model, reducing the number of parameters and computational resources. Finally, the SGD optimizer and GIOU loss function, as well as the Sim attention module, were used to improve detection accuracy. Based on instance analysis and comparison, this method achieves [email protected] of 96.9 % and detection speed of 66FPS. Compared with YOLOv5s, it reduces the number of parameters by 79 % and FLOPs by 77 %.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105830"},"PeriodicalIF":9.6,"publicationDate":"2024-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534918","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}