Scholarlink

<p>To overcome the inefficiency, high cost, and limited real-time capability of conventional building renovation modeling methods, this study proposed a dynamic modeling framework integrating Internet of Things (IoT) sensing and unmanned aerial vehicle (UAV) vision. An IoT monitoring system based on micro-electro-mechanical system (MEMS) accelerometers was developed, incorporating an LTE-Cat1 communication module to enable high-sensitivity, long-term vibration data acquisition and transmission. A microcontroller-based preprocessing strategy was implemented to enhance data accuracy and real-time responsiveness. A unified IoT platform was further established for multi-source data management and structural dynamic analysis. For geometric and surface condition acquisition, UAV oblique photogrammetry was employed to capture high-precision facade and structural detail data. A crack detection framework based on You Only Look Once v5–Dense Enhanced (YOLOv5-DE) was designed by introducing a feature enhancement module that jointly exploited low- and high-dimensional features. Through a densely connected feature fusion mechanism and EFConv-based optimization, the model achieved improved robustness under complex scene conditions while maintaining lightweight characteristics. Experiments conducted on the Crack-2218 dataset demonstrated that the proposed network achieved a detection accuracy of 96.5%, with only 1.4 million parameters and an inference time of 3.35&#xa0;ms. Compared with traditional renovation modeling approaches relying on manual inspection and sparse sensing, the proposed framework increased data acquisition efficiency by approximately 80%, reduced modeling time by 75%, and improved crack detection accuracy from 80 to 96.5%. Ablation studies further verified the effectiveness of the Dense Enhanced module, which reduced weight size by 46.9% and total parameters by 53.1%. The proposed approach overcame the limitations of single-source monitoring and enabled the transition from static assessment to dynamic structural evaluation, providing a practical and scalable technical pathway for intelligent building renovation modeling.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluation of building structural reinforcement performance by internet of things and UAV sensing

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月29日

查看全文
  • Weiguo Yang,
  • Feng Jiang,
  • Zhizhao Shao,
  • Yinan Qi,
  • Longxiang Wan

摘要

To overcome the inefficiency, high cost, and limited real-time capability of conventional building renovation modeling methods, this study proposed a dynamic modeling framework integrating Internet of Things (IoT) sensing and unmanned aerial vehicle (UAV) vision. An IoT monitoring system based on micro-electro-mechanical system (MEMS) accelerometers was developed, incorporating an LTE-Cat1 communication module to enable high-sensitivity, long-term vibration data acquisition and transmission. A microcontroller-based preprocessing strategy was implemented to enhance data accuracy and real-time responsiveness. A unified IoT platform was further established for multi-source data management and structural dynamic analysis. For geometric and surface condition acquisition, UAV oblique photogrammetry was employed to capture high-precision facade and structural detail data. A crack detection framework based on You Only Look Once v5–Dense Enhanced (YOLOv5-DE) was designed by introducing a feature enhancement module that jointly exploited low- and high-dimensional features. Through a densely connected feature fusion mechanism and EFConv-based optimization, the model achieved improved robustness under complex scene conditions while maintaining lightweight characteristics. Experiments conducted on the Crack-2218 dataset demonstrated that the proposed network achieved a detection accuracy of 96.5%, with only 1.4 million parameters and an inference time of 3.35 ms. Compared with traditional renovation modeling approaches relying on manual inspection and sparse sensing, the proposed framework increased data acquisition efficiency by approximately 80%, reduced modeling time by 75%, and improved crack detection accuracy from 80 to 96.5%. Ablation studies further verified the effectiveness of the Dense Enhanced module, which reduced weight size by 46.9% and total parameters by 53.1%. The proposed approach overcame the limitations of single-source monitoring and enabled the transition from static assessment to dynamic structural evaluation, providing a practical and scalable technical pathway for intelligent building renovation modeling.