Automation in Construction最新文献

英文中文

Structural defect segmentation using a semi-supervised algorithm integrating YOLO and the segment anything model 结合YOLO和任意分割模型的半监督结构缺陷分割算法

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-11 DOI: 10.1016/j.autcon.2025.106709

Gauthami Vijayakumar Kuttuva, Prawin J

Deep learning–based defect segmentation enables automated detection, classification, localisation, and quantification of structural defects. Although supervised segmentation methods yield strong performance, they require extensive pixel-level annotations and struggle with data scarcity, class imbalance, thin crack segmentation, and generalisation across diverse conditions. This paper proposes a semi-supervised two-stage segmentation framework that integrates a high-accuracy object detection model (Stage I: localisation using YOLO11/Oriented YOLO11) with zero-shot unsupervised segmentation (Stage II: pixel-level mapping using Segment Anything Model with box prompts). The method requires only object-detection labels, eliminating the need for dedicated segmentation annotations. Experiments on benchmark datasets involving steel, masonry, concrete cracks, spalling, and corrosion demonstrate that the hybrid Oriented YOLO11 and SAM model achieves state-of-the-art performance, with an average Dice score of 0.7, comparable to those of fully supervised models. The proposed framework offers real-time performance, strong generalisation, and high scalability, making it a promising solution for robust structural defect segmentation.

基于深度学习的缺陷分割实现了结构缺陷的自动检测、分类、定位和量化。尽管监督分割方法产生了强大的性能，但它们需要大量的像素级注释，并与数据稀缺性、类不平衡、细裂纹分割和不同条件下的泛化作斗争。本文提出了一种半监督两阶段分割框架，该框架集成了高精度目标检测模型（第一阶段：使用YOLO11/Oriented YOLO11进行定位）和零镜头无监督分割（第二阶段：使用带有框提示的Segment Anything model进行像素级映射）。该方法只需要对象检测标签，不需要专门的分割注释。在涉及钢铁、砌体、混凝土裂缝、剥落和腐蚀的基准数据集上的实验表明，混合的面向YOLO11和SAM模型达到了最先进的性能，平均Dice得分为0.7，与完全监督模型相当。该框架具有实时性好、泛化能力强、可扩展性强等特点，是鲁棒结构缺陷分割的理想解决方案。

{"title":"Structural defect segmentation using a semi-supervised algorithm integrating YOLO and the segment anything model","authors":"Gauthami Vijayakumar Kuttuva, Prawin J","doi":"10.1016/j.autcon.2025.106709","DOIUrl":"10.1016/j.autcon.2025.106709","url":null,"abstract":"<div><div>Deep learning–based defect segmentation enables automated detection, classification, localisation, and quantification of structural defects. Although supervised segmentation methods yield strong performance, they require extensive pixel-level annotations and struggle with data scarcity, class imbalance, thin crack segmentation, and generalisation across diverse conditions. This paper proposes a semi-supervised two-stage segmentation framework that integrates a high-accuracy object detection model (Stage I: localisation using YOLO11/Oriented YOLO11) with zero-shot unsupervised segmentation (Stage II: pixel-level mapping using Segment Anything Model with box prompts). The method requires only object-detection labels, eliminating the need for dedicated segmentation annotations. Experiments on benchmark datasets involving steel, masonry, concrete cracks, spalling, and corrosion demonstrate that the hybrid Oriented YOLO11 and SAM model achieves state-of-the-art performance, with an average Dice score of 0.7, comparable to those of fully supervised models. The proposed framework offers real-time performance, strong generalisation, and high scalability, making it a promising solution for robust structural defect segmentation.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106709"},"PeriodicalIF":11.5,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145734739","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}

引用次数: 0

Automated and scalable BIM2BEM framework with zoning-based model simplification leveraging knowledge graph integration 自动化和可扩展的BIM2BEM框架，基于分区的模型简化利用知识图集成

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-11 DOI: 10.1016/j.autcon.2025.106712

Meng Wang, Georgios N. Lilis, Dimitris Mavrokapnidis, Kyriakos Katsigarakis, Ivan Korolija, Dimitrios Rovas

Accurate and scalable generation of Building Energy Models (BEM) from Building Information Modelling (BIM) data is critical for performance-driven building design. However, existing methods are often constrained by data quality issues and rigid workflows, limiting automation. This paper proposes an automated and scalable BIM-to-BEM (BIM2BEM) framework enabled by knowledge graph integration, designed to support automation and scalability in model generation from imperfect BIM data. To manage model complexity, zoning-based mappings from BIM spaces to thermal zones are derived through multi-factor analysis of spatial relationships, functional usage, thermal load similarity, and HVAC configuration. Applied to a real-world complex building, the framework reduces simulation time by up to 70%, while maintaining energy use deviations within 3% and HVAC sizing variations up to 10%, compared with the full-model baseline. These findings indicate that the proposed framework can enhance BIM2BEM automation, supporting the scalable and flexible generation of simulation-ready models under practical data limitations.

从建筑信息模型（BIM）数据中准确和可扩展地生成建筑能源模型（BEM）对于性能驱动型建筑设计至关重要。然而，现有的方法经常受到数据质量问题和严格的工作流程的限制，限制了自动化。本文提出了一个自动化和可扩展的BIM-to- bem （BIM2BEM）框架，该框架通过知识图集成实现，旨在支持从不完善的BIM数据生成模型的自动化和可扩展性。为了管理模型的复杂性，通过对空间关系、功能使用、热负荷相似性和暖通空调配置的多因素分析，得出了从BIM空间到热区基于分区的映射。应用于现实世界的复杂建筑，与全模型基线相比，该框架将模拟时间缩短了70%，同时将能源使用偏差保持在3%以内，暖通空调尺寸变化保持在10%以内。这些发现表明，所提出的框架可以增强BIM2BEM自动化，支持在实际数据限制下可扩展和灵活地生成仿真就绪模型。

{"title":"Automated and scalable BIM2BEM framework with zoning-based model simplification leveraging knowledge graph integration","authors":"Meng Wang, Georgios N. Lilis, Dimitris Mavrokapnidis, Kyriakos Katsigarakis, Ivan Korolija, Dimitrios Rovas","doi":"10.1016/j.autcon.2025.106712","DOIUrl":"10.1016/j.autcon.2025.106712","url":null,"abstract":"<div><div>Accurate and scalable generation of Building Energy Models (BEM) from Building Information Modelling (BIM) data is critical for performance-driven building design. However, existing methods are often constrained by data quality issues and rigid workflows, limiting automation. This paper proposes an automated and scalable BIM-to-BEM (BIM2BEM) framework enabled by knowledge graph integration, designed to support automation and scalability in model generation from imperfect BIM data. To manage model complexity, zoning-based mappings from BIM spaces to thermal zones are derived through multi-factor analysis of spatial relationships, functional usage, thermal load similarity, and HVAC configuration. Applied to a real-world complex building, the framework reduces simulation time by up to 70%, while maintaining energy use deviations within 3% and HVAC sizing variations up to 10%, compared with the full-model baseline. These findings indicate that the proposed framework can enhance BIM2BEM automation, supporting the scalable and flexible generation of simulation-ready models under practical data limitations.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106712"},"PeriodicalIF":11.5,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145732333","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}

引用次数: 0

Diffusion-enhanced semantic segmentation for underground crack detection 基于扩散增强语义分割的地下裂缝检测

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-11 DOI: 10.1016/j.autcon.2025.106721

Neng Wang , Bowen Jiang , Camillo J. Taylor , Zili Li

Unlike surface structures, crack detection in underground environments faces greater challenges due to low illumination and complex structural context. This paper develops a diffusion-based model for crack segmentation in underground tunnel structures. Building on a framework that reformulates semantic segmentation as conditional image generation, an enhanced diffusion architecture guided by prior masks is proposed. Trained on a diverse dataset comprising crack images from the Dublin Port Tunnel, the European Organization for Nuclear Research, and the Bochum Crack Dataset, the proposed model outperforms benchmark models (e.g., Mask2Former), achieving a 9.1% increase in mIoU and 16.1% in F1-score in multi-scenario validation. In field testing, the proposed two-stage pipeline optimizes inference strategy by excluding crack-free patches (87.5% of the total), and tests real-world applicability, with an absolute mIoU gain of 0.063. Additionally, a post-processing strategy leveraging uncertainty maps further refines the segmentation results.

与地面结构不同，由于光照不足和复杂的结构环境，地下环境的裂缝检测面临着更大的挑战。本文建立了一种基于扩散的地下隧道结构裂缝分割模型。在将语义分割作为条件图像生成的框架的基础上，提出了一种基于先验掩码的增强扩散架构。在都柏林港口隧道、欧洲核研究组织和波鸿裂缝数据集的裂缝图像的不同数据集上进行训练，所提出的模型优于基准模型（例如Mask2Former），在多场景验证中实现了9.1%的mIoU和16.1%的f1分数的提高。在现场测试中，提出的两阶段管道通过排除无裂纹补丁（占总数的87.5%）来优化推理策略，并测试了现实世界的适用性，绝对mIoU增益为0.063。此外，利用不确定性映射的后处理策略进一步细化了分割结果。

引用次数: 0

From conventional brain–computer interfaces to gen-AI–powered systems for advancing construction safety management 从传统的脑机接口到推进建筑安全管理的人工智能驱动系统

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-11 DOI: 10.1016/j.autcon.2025.106708

Ah Khoon Hwee , Tan Kah Hui , Liu QingHua , Wang Jiao

Although the use of brain–computer interfaces (BCI) in construction safety management is expanding, no comprehensive review has yet examined how generative artificial intelligence (Gen-AI) can enhance BCI systems in this domain. A total of 292 peer-reviewed journal articles are analysed in this review. The state-of-the-art applications such as simulation, real-time monitoring, and human-robot collaboration and their associated limitations are reviewed and scrutinised. Unlike previous BCI safety reviews, this paper highlights the integrative role of Gen-AI in enhancing BCI systems through noise reduction, data augmentation, and self-supervised learning frameworks. Furthermore, a visual roadmap outlining short-, medium-, and long-term is proposed based on the challenges identified in Gen-AI enhanced BCI systems. The findings are expected to assist researchers in understanding the current state of BCI development and to provide a perspective from which future directions can be formulated to advance construction safety management.

尽管脑机接口（BCI）在建筑安全管理中的应用正在扩大，但尚未有全面的综述研究生成式人工智能（Gen-AI）如何增强该领域的BCI系统。本综述共分析了292篇同行评议的期刊文章。最先进的应用，如仿真，实时监控，人机协作及其相关的限制进行了审查和审查。与以往的脑机接口安全性综述不同，本文强调了Gen-AI通过降噪、数据增强和自我监督学习框架在增强脑机接口系统方面的综合作用。此外，根据Gen-AI增强BCI系统中确定的挑战，提出了一份概述短期、中期和长期的可视化路线图。研究结果有望帮助研究人员了解脑机接口的发展现状，并为制定未来的发展方向提供一个视角，以推进建筑安全管理。

引用次数: 0

Real-time on-site structural safety assessment of metro tunnel linings via WSN and edge computing 基于无线传感器网络和边缘计算的地铁隧道衬砌结构安全实时现场评估

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-10 DOI: 10.1016/j.autcon.2025.106711

Dongming Zhang , Jianhui Yu , Mingliang Zhou , Hongwei Huang , Linghan Ouyang

Edge computing mitigates the latency of traditional “on-site acquisition–off-site processing” workflows, enabling real-time structural safety assessment. This paper develops an edge-computing-based safety assessment system for shield tunnel linings, integrating sensors, inspection technologies, wireless sensor networks, and edge gateways. Coordinated gateway–sensor communication enables on-site data fusion and standardized processing for variable-weight fuzzy assessment. A lookup-table-based membership function achieved a 79× acceleration, 74 % lower memory use, and 34 % lower energy consumption. A lightweight multithreaded architecture improved image pre-processing by 60 %, while an optimized Kalman filter reduced latency and energy by 20 % and 36 %, respectively. A simplified Seasonal-Trend decomposition using Loess (STL)-ARIMA model enhanced forecasting efficiency by 14 %. Validated on Shanghai Metro Line 2 in China, the system enabled zone-level assessments within 243 s. By integrating edge-compatible algorithms with domain-specific structural knowledge, the framework provides a scalable, energy-efficient, and adaptive solution for long-term intelligent maintenance of shield tunnels and similar infrastructure systems.

边缘计算减少了传统“现场采集-非现场处理”工作流程的延迟，实现了实时结构安全评估。本文开发了一种基于边缘计算的盾构隧道衬砌安全评估系统，该系统集成了传感器、检测技术、无线传感器网络和边缘网关。网关-传感器协同通信，实现现场数据融合和标准化处理，实现变权模糊评价。基于查询表的成员函数实现了79x的加速、74%的内存使用和34%的能耗。轻量级多线程架构将图像预处理提高了60%，而优化的卡尔曼滤波器将延迟和能量分别降低了20%和36%。采用黄土(STL)-ARIMA模型简化季节趋势分解，预报效率提高14%。该系统在中国上海地铁2号线进行了验证，在243秒内实现了区域级评估。通过将边缘兼容算法与特定领域的结构知识相结合，该框架为盾构隧道和类似基础设施系统的长期智能维护提供了可扩展、节能和自适应的解决方案。

{"title":"Real-time on-site structural safety assessment of metro tunnel linings via WSN and edge computing","authors":"Dongming Zhang , Jianhui Yu , Mingliang Zhou , Hongwei Huang , Linghan Ouyang","doi":"10.1016/j.autcon.2025.106711","DOIUrl":"10.1016/j.autcon.2025.106711","url":null,"abstract":"<div><div>Edge computing mitigates the latency of traditional “on-site acquisition–off-site processing” workflows, enabling real-time structural safety assessment. This paper develops an edge-computing-based safety assessment system for shield tunnel linings, integrating sensors, inspection technologies, wireless sensor networks, and edge gateways. Coordinated gateway–sensor communication enables on-site data fusion and standardized processing for variable-weight fuzzy assessment. A lookup-table-based membership function achieved a 79× acceleration, 74 % lower memory use, and 34 % lower energy consumption. A lightweight multithreaded architecture improved image pre-processing by 60 %, while an optimized Kalman filter reduced latency and energy by 20 % and 36 %, respectively. A simplified Seasonal-Trend decomposition using Loess (STL)-ARIMA model enhanced forecasting efficiency by 14 %. Validated on Shanghai Metro Line 2 in China, the system enabled zone-level assessments within 243 s. By integrating edge-compatible algorithms with domain-specific structural knowledge, the framework provides a scalable, energy-efficient, and adaptive solution for long-term intelligent maintenance of shield tunnels and similar infrastructure systems.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106711"},"PeriodicalIF":11.5,"publicationDate":"2025-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145734741","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}

引用次数: 0

Generalizable deep sequence models for 4D trajectory prediction of tower crane loads 塔机载荷四维轨迹预测的广义深度序列模型

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-10 DOI: 10.1016/j.autcon.2025.106696

Mohammad Hossein Kazemi , Yuqing Hu , Yi Wu , John I. Messner , Scarlett R. Miller

Predicting the trajectory of suspended loads is essential for proactive collision avoidance and improving situational awareness during tower crane operations. Existing approaches suffer from limited generalizability, simplified motion assumptions, and lack of real-time deployment. This paper presents a data-driven framework for 4D (3D + time) trajectory prediction using deep sequence models trained on realistic crane-operation data. A Unity-based simulation environment is developed to emulate rotary and linear encoders, and 29 participants operate the crane across randomized pick-and-place tasks, producing diverse motion trajectories. Six architectures — including LSTM-based Seq2Seq models with different attention mechanisms, ConvLSTM networks, and Temporal Convolutional Networks — under varying prediction horizons, temporal context, sampling rates, and sensor noise are evaluated. The Seq2Seq model with Temporal Attention achieves the best performance, with a mean 3D displacement error of 0.45 m on unseen logistic scenarios. A high-performing model is integrated into a real-time digital twin to provide feedback for operator training.

在塔吊运行过程中，预测吊载轨迹对于主动避免碰撞和提高态势感知能力至关重要。现有的方法受限于有限的通用性，简化的运动假设，以及缺乏实时部署。本文提出了一种数据驱动的四维（3D +时间）轨迹预测框架，该框架采用基于实际起重机操作数据训练的深度序列模型。开发了一个基于unity的仿真环境来模拟旋转和线性编码器，29名参与者在随机拾取和放置任务中操作起重机，产生不同的运动轨迹。在不同的预测范围、时间背景、采样率和传感器噪声下，评估了六种架构，包括基于lstm的具有不同注意机制的Seq2Seq模型、ConvLSTM网络和Temporal Convolutional networks。具有时间注意力的Seq2Seq模型达到了最好的性能，在不可见的物流场景下平均3D位移误差为0.45 m。将高性能模型集成到实时数字孪生模型中，为操作员培训提供反馈。

{"title":"Generalizable deep sequence models for 4D trajectory prediction of tower crane loads","authors":"Mohammad Hossein Kazemi , Yuqing Hu , Yi Wu , John I. Messner , Scarlett R. Miller","doi":"10.1016/j.autcon.2025.106696","DOIUrl":"10.1016/j.autcon.2025.106696","url":null,"abstract":"<div><div>Predicting the trajectory of suspended loads is essential for proactive collision avoidance and improving situational awareness during tower crane operations. Existing approaches suffer from limited generalizability, simplified motion assumptions, and lack of real-time deployment. This paper presents a data-driven framework for 4D (3D + time) trajectory prediction using deep sequence models trained on realistic crane-operation data. A Unity-based simulation environment is developed to emulate rotary and linear encoders, and 29 participants operate the crane across randomized pick-and-place tasks, producing diverse motion trajectories. Six architectures — including LSTM-based Seq2Seq models with different attention mechanisms, ConvLSTM networks, and Temporal Convolutional Networks — under varying prediction horizons, temporal context, sampling rates, and sensor noise are evaluated. The Seq2Seq model with Temporal Attention achieves the best performance, with a mean 3D displacement error of 0.45 m on unseen logistic scenarios. A high-performing model is integrated into a real-time digital twin to provide feedback for operator training.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106696"},"PeriodicalIF":11.5,"publicationDate":"2025-12-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145730976","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}

引用次数: 0

Design, multi-scale structural analysis, and construction of modular prefabricated 3D-printed concrete residence 模块化预制3d打印混凝土住宅的设计、多尺度结构分析和施工

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-09 DOI: 10.1016/j.autcon.2025.106705

Hailong Wang, Yiqing Shi, Xiaoyan Sun, Xiqiang Lin, Yinlin Ye

Residential building construction is increasingly challenged by labour shortages and limited funding. Modular prefabricated 3D printing concrete (3DPC) technology, integrated with digital design and automated construction, offers an efficient and standardised solution to the construction of residential buildings, and has seen emerging applications worldwide. This paper presents the process from architectural and structural design based on one digital model to 3D printing and assembly of a modular prefabricated 3DPC residence, incorporating a multi-scale numerical simulation method for structural analysis. The study identifies the primary limitation of modular 3DPC as the size constraints imposed by printing and transportation equipment. A case study in Hebei Province, China, demonstrates the practicality. Results indicate that under the most unfavourable conditions, the maximum stresses are within 1.3 % of the limit values, ensuring a high safety margin. The 3DPC residence demonstrates time and cost savings, with digital design optimizing coordination among structural, architectural, and MEP systems.

住房建设日益受到劳动力短缺和资金有限的挑战。模块化预制3D打印混凝土（3DPC）技术与数字设计和自动化施工相结合，为住宅建筑的建造提供了高效、标准化的解决方案，并在全球范围内得到了广泛的应用。本文介绍了模块化预制3DPC住宅从基于一个数字模型的建筑和结构设计到3D打印和组装的过程，并采用多尺度数值模拟方法进行结构分析。该研究确定了模块化3DPC的主要限制是打印和运输设备对尺寸的限制。以河北省为例，验证了该方法的实用性。结果表明，在最不利的条件下，最大应力在限值的1.3%以内，保证了较高的安全裕度。3DPC住宅展示了时间和成本的节省，数字化设计优化了结构、建筑和MEP系统之间的协调。

{"title":"Design, multi-scale structural analysis, and construction of modular prefabricated 3D-printed concrete residence","authors":"Hailong Wang, Yiqing Shi, Xiaoyan Sun, Xiqiang Lin, Yinlin Ye","doi":"10.1016/j.autcon.2025.106705","DOIUrl":"10.1016/j.autcon.2025.106705","url":null,"abstract":"<div><div>Residential building construction is increasingly challenged by labour shortages and limited funding. Modular prefabricated 3D printing concrete (3DPC) technology, integrated with digital design and automated construction, offers an efficient and standardised solution to the construction of residential buildings, and has seen emerging applications worldwide. This paper presents the process from architectural and structural design based on one digital model to 3D printing and assembly of a modular prefabricated 3DPC residence, incorporating a multi-scale numerical simulation method for structural analysis. The study identifies the primary limitation of modular 3DPC as the size constraints imposed by printing and transportation equipment. A case study in Hebei Province, China, demonstrates the practicality. Results indicate that under the most unfavourable conditions, the maximum stresses are within 1.3 % of the limit values, ensuring a high safety margin. The 3DPC residence demonstrates time and cost savings, with digital design optimizing coordination among structural, architectural, and MEP systems.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106705"},"PeriodicalIF":11.5,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145734740","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}

引用次数: 0

Lightweight railway defect detection model with attention-based feature fusion and GPS mapping 基于注意力特征融合和GPS映射的轻型铁路缺陷检测模型

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-09 DOI: 10.1016/j.autcon.2025.106704

Minh-Tu Cao , Wei-Chih Wang , Michael Koean

Automatic and accurate identification of rail defects is critical to reducing the risk of accidents and ensuring service reliability. This paper introduces YOLO-TDR, an advanced YOLO-based model for railway defect detection. YOLO-TDR integrates Rep-Net with Cross-Stage Partial and Efficient Layer Aggregation Networks, and Cross-level Convolution Fusion Polarized Self-Attention modules to enhance feature extraction and attention-guided fusion while the WiseIoUv3 loss improves localization precision. Evaluated on a real-world dataset of 9367 images with 71,632 defect instances, YOLO-TDR achieved the greatest mAP@50 and mAP@50–95 values (88.4 % and 69.9 % for three-class) and (94.6 % and 86.1 % for seven-class). YOLO-TDR outperformed YOLOv8-v11 models at least 0.7 % mAP@50–95 for the seven-class test and 2.1 % for broken sleeper detection particularly. Combined with GPS-based georeferencing and a tracking technique, YOLO-TDR enables real-time mapping of railway defect localization along the inspection path. These results demonstrate significant contributions of YOLO-TDR to algorithmic development and the advancement of railway inspection automation.

钢轨缺陷的自动准确识别对于降低事故风险和确保服务可靠性至关重要。介绍了一种先进的基于yolo的铁路缺陷检测模型YOLO-TDR。YOLO-TDR将Rep-Net与Cross-Stage Partial和Efficient Layer Aggregation Networks、Cross-level Convolution Fusion极化自注意模块集成在一起，增强了特征提取和注意引导融合，而WiseIoUv3 loss则提高了定位精度。在包含9367张图像和71632个缺陷实例的真实数据集上进行评估，YOLO-TDR获得了最大的mAP@50和mAP@50 -95值（3类为88.4%和69.9%）和（7类为94.6%和86.1%）。YOLO-TDR在七类测试中优于YOLOv8-v11模型至少0.7% mAP@50 -95，特别是在枕套破损检测中优于2.1%。结合基于gps的地理参考和跟踪技术，YOLO-TDR可以沿着检查路径实时绘制铁路缺陷定位图。这些结果表明，YOLO-TDR对算法的发展和铁路检测自动化的进步做出了重大贡献。

{"title":"Lightweight railway defect detection model with attention-based feature fusion and GPS mapping","authors":"Minh-Tu Cao , Wei-Chih Wang , Michael Koean","doi":"10.1016/j.autcon.2025.106704","DOIUrl":"10.1016/j.autcon.2025.106704","url":null,"abstract":"<div><div>Automatic and accurate identification of rail defects is critical to reducing the risk of accidents and ensuring service reliability. This paper introduces YOLO-TDR, an advanced YOLO-based model for railway defect detection. YOLO-TDR integrates Rep-Net with Cross-Stage Partial and Efficient Layer Aggregation Networks, and Cross-level Convolution Fusion Polarized Self-Attention modules to enhance feature extraction and attention-guided fusion while the WiseIoUv3 loss improves localization precision. Evaluated on a real-world dataset of 9367 images with 71,632 defect instances, YOLO-TDR achieved the greatest mAP@50 and mAP@50<em>–</em>95 values (88.<em>4 %</em> and 69.<em>9 %</em> for three-class) and (94.<em>6 %</em> and 86.<em>1 %</em> for seven-class). YOLO-TDR outperformed YOLOv8-v11 models at least 0.<em>7 %</em> mAP@50<em>–</em>95 for the seven-class test and 2.<em>1 %</em> for broken sleeper detection particularly. Combined with GPS-based georeferencing and a tracking technique, YOLO-TDR enables real-time mapping of railway defect localization along the inspection path. These results demonstrate significant contributions of YOLO-TDR to algorithmic development and the advancement of railway inspection automation.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"182 ","pages":"Article 106704"},"PeriodicalIF":11.5,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145734742","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}

引用次数: 0

BIM-driven digital risk twins for tunnel reinforcement maintenance bim驱动的隧道加固维护数字风险孪生

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-08 DOI: 10.1016/j.autcon.2025.106710

Junhwi Cho , Junseo Lee , Byung-Dal So , Jae-Hyun Kim , Jung-Doung Yu , Jaeheum Yeon

Maintaining the structural integrity of tunnels is crucial, yet assessing embedded reinforcements such as rock bolts remains difficult with conventional methods. This paper proposes a digital risk twin (DRT) framework to operationalize risk-based maintenance (RBM). Unlike conventional digital twins (DT) that focus on state visualization, the DRT visualizes risk as the primary output by converting rock-bolt measurements into failure-mode risk metrics with uncertainty propagation and rendering a tiered 3D risk map. The framework couples a Building Information Model (BIM) with sensor streams. Through visual programming, strain data are automatically mapped to the model and color-coded by risk which enables rapid localization and prioritization of at-risk areas. Laboratory tests on instrumented rock bolts validate the end-to-end reliability of the system. The proposed approach is expected to provide a transparent and efficient basis for RBM decisions, reducing uncertainty and minimizing risk and life cycle costs.

保持隧道结构的完整性是至关重要的，然而，用传统方法评估岩石螺栓等嵌入式加固仍然很困难。本文提出了一个数字风险孪生（DRT）框架来实现基于风险的维护（RBM）。与专注于状态可视化的传统数字孪生（DT）不同，DRT通过将锚杆测量值转换为具有不确定性传播的失效模式风险指标，并呈现分层的3D风险图，将风险可视化为主要输出。该框架将建筑信息模型（BIM）与传感器流耦合在一起。通过可视化编程，应变数据自动映射到模型中，并根据风险进行颜色编码，从而能够快速定位和确定风险区域的优先级。对锚杆仪器的实验室测试验证了该系统的端到端可靠性。预计拟议的方法将为RBM决策提供透明和有效的基础，减少不确定性并最大限度地降低风险和生命周期成本。

引用次数: 0

Predicting building layout structure and features via planar duality and graph neural networks 利用平面对偶和图神经网络预测建筑布局结构和特征

IF 11.5 1区工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY

Automation in Construction

Pub Date : 2025-12-08 DOI: 10.1016/j.autcon.2025.106681

Kuntao Hu , Nannan Zhang , Tianning Yao , Ziqi Xu , Xing Chen , Liang Sun

Predicting spatial topology and feature parameters during architectural design is crucial for AI-driven automation. This paper introduces planar duality theory, employing a “node-edge-face” graph representation and an innovative face-adjacency-based message passing mechanism to replace traditional edge-based propagation. The proposed DualGAT model is evaluated on 48 floorplans from nine educational buildings. DualGAT effectively generates complex layouts with 21-dimensional features, achieving 0.836 accuracy (±10 % tolerance) for point-feature prediction and 0.774 accuracy for edge-connection prediction, and significantly outperforming a standard Graph Attention Network (GAT) with equivalent parameters. The results confirm the necessity of incorporating “face” elements and developing tailor-made AI models like DualGAT within graph representations to enhance predictive accuracy for design-phase building automation.

在建筑设计过程中预测空间拓扑和特征参数对于人工智能驱动的自动化至关重要。本文引入平面对偶理论，采用“节点-边-面”图表示和创新的基于面邻接的消息传递机制来取代传统的基于边的传播。提出的DualGAT模型对来自9个教育建筑的48个平面图进行了评估。DualGAT有效地生成具有21维特征的复杂布局，点特征预测精度达到0.836（±10%公差），边缘连接预测精度达到0.774，显著优于具有等效参数的标准图注意网络（GAT）。结果证实了在图形表示中纳入“人脸”元素和开发量身定制的AI模型（如DualGAT）的必要性，以提高设计阶段建筑自动化的预测准确性。

引用次数: 0

首页上一页

下一页尾页

类型

全部化学•材料生命科学医学物理工程技术环境•农林材料科学地球科学法学管理学化学环境科学与生态学计算机科学教育学经济学农林科学人文科学生物学数学物理与天体物理心理学综合性期刊其他工业工程理学历史学农学文学信息工程

数据库

全部 ACS Publications Elsevier ieeexplore Springer The Royal Society of Chemistry Wiley

期刊

Automation in Construction

全部 Acc. Chem. Res. ACS Applied Bio Materials ACS Appl. Electron. Mater. ACS Appl. Energy Mater. ACS Appl. Mater. Interfaces ACS Appl. Nano Mater. ACS Appl. Polym. Mater. ACS BIOMATER-SCI ENG ACS Catal. ACS Cent. Sci. ACS Chem. Biol. ACS Chemical Health & Safety ACS Chem. Neurosci. ACS Comb. Sci. ACS Earth Space Chem. ACS Energy Lett. ACS Infect. Dis. ACS Macro Lett. ACS Mater. Lett. ACS Med. Chem. Lett. ACS Nano ACS Omega ACS Photonics ACS Sens. ACS Sustainable Chem. Eng. ACS Synth. Biol. Anal. Chem. BIOCHEMISTRY-US Bioconjugate Chem. BIOMACROMOLECULES Chem. Res. Toxicol. Chem. Rev. Chem. Mater. CRYST GROWTH DES ENERG FUEL Environ. Sci. Technol. Environ. Sci. Technol. Lett. Eur. J. Inorg. Chem. IND ENG CHEM RES Inorg. Chem. J. Agric. Food. Chem. J. Chem. Eng. Data J. Chem. Educ. J. Chem. Inf. Model. J. Chem. Theory Comput. J. Med. Chem. J. Nat. Prod. J PROTEOME RES J. Am. Chem. Soc. LANGMUIR MACROMOLECULES Mol. Pharmaceutics Nano Lett. Org. Lett. ORG PROCESS RES DEV ORGANOMETALLICS J. Org. Chem. J. Phys. Chem. J. Phys. Chem. A J. Phys. Chem. B J. Phys. Chem. C J. Phys. Chem. Lett. Analyst Anal. Methods Biomater. Sci. Catal. Sci. Technol. Chem. Commun. Chem. Soc. Rev. CHEM EDUC RES PRACT CRYSTENGCOMM Dalton Trans. Energy Environ. Sci. ENVIRON SCI-NANO ENVIRON SCI-PROC IMP ENVIRON SCI-WAT RES Faraday Discuss. Food Funct. Green Chem. Inorg. Chem. Front. Integr. Biol. J. Anal. At. Spectrom. J. Mater. Chem. A J. Mater. Chem. B J. Mater. Chem. C Lab Chip Mater. Chem. Front. Mater. Horiz. MEDCHEMCOMM Metallomics Mol. Biosyst. Mol. Syst. Des. Eng. Nanoscale Nanoscale Horiz. Nat. Prod. Rep. New J. Chem. Org. Biomol. Chem. Org. Chem. Front. PHOTOCH PHOTOBIO SCI PCCP Polym. Chem.

﹀