摘要
随着智慧工地建设的快速推进，AI巡检系统成为保障施工安全和效率的重要工具。然而，误报率过高不仅影响系统的可靠性，也增加了人工复核成本，制约了其广泛应用。本文聚焦于智慧工地AI巡检系统的误报率问题，分析其成因，结合多源数据融合、深度学习模型优化及阈值动态调整等策略，有效降低误报率，提高系统的准确性和稳定性，为智慧工地安全管理提供技术支撑。
关键词： 智慧工地；AI巡检；误报率；深度学习；数据融合
Abstract
As the rapid advancement of smart construction sites progresses, AI inspection systems have become a crucial tool for ensuring construction safety and efficiency. However, high false alarm rates not only affect system reliability but also increase the cost of manual verification, limiting their widespread application. This paper focuses on the issue of false alarm rates in AI inspection systems for smart construction sites, analyzing its causes. By integrating multi-source data fusion, optimizing deep learning models, and dynamically adjusting thresholds, we aim to effectively reduce false alarm rates, enhance system accuracy and stability, and provide technical support for the safety management of smart construction sites.
Key words： Smart construction site; AI inspection; False alarm rate; Deep learning; Data fusion
参考文献 References
[1] 赵龙.基于智慧工地系统的建筑施工现场管理策略[J].城市建设理论研究(电子版),2025,(14):38-40.
[2] 王通,黄修平,武齐齐,等.智慧工地场景下桥梁基础大体积混凝土温控系统应用研究[J/OL].公路,2025,(05):92-101[2025-05-27].
[3] 张善松.智慧工地建设背景下住宅建筑工程安全管理措施[J].居舍,2025,(13):169-172.
[4] 曹旭.机器狗AI技术在铁路线路巡检中的应用[J].中国科技信息,2025,(09):85-87.
[5] 孔淑梅,周国贞.基于数字孪生和AI视觉识别技术的加热炉智能巡检系统的研究[J].南方金属,2025,(02):73-75+81.
[6] 田洪伟,撒兴才,李钰,等.基于联动控制与AI视觉识别技术的地下管廊智能巡检机器人研究[J].科技创新与应用,2025,15(10):10-13.
[7] 郭缤璐,张天元.AI店小二DeepSeek智能巡检餐饮业迎人工智能3.0[N].北京商报,2025-04-01(004).
[8] 罗文延,李林波,赵赫,等.基于AI技术的无人机光伏巡检系统构建与优化[J].光源与照明,2025,(03):172-174.