整合物聯網感測與人工智慧於智慧型災防系統之研究

Abstract

本研究旨在建構一套整合物聯網感測技術、先進通訊介面與人工智慧模型之智慧型環境災防系統。系統涵蓋多模組感測裝置，包括人工坡地監測模組、氣象盒子、水利智慧粒子、空氣盒子與智慧水井，能即時監測微氣象、水文參數、有害氣體與空氣品質及物件相對變形量等多元環境資訊，並結合GNSS定位、三軸加速度計與影像辨識技術，以提升異常事件偵測的精度與空間解析度。進一步，本系統融合大型語言模型(Large Language Models, LLMs)與通訊平台(如LINE與Messenger)，實現防災資訊的自動化傳遞、即時互動與個人化溝通，強化社區層級的風險感知與應變能力。透過實地部署與多場域測試驗證，本系統展現出優異的擴充彈性與即時反應效能，能有效支援區域性災害預警與風險傳遞，具高度應用潛力，有助於建構具韌性的智慧城市與社區。

This study presents the development of an intelligent environmental disaster prevention system that integrates Internet of Things (IoT) sensing technologies, advanced communication interfaces, and artificial intelligence models. The proposed system comprises a suite of modular sensing units, including slope monitoring modules, weather boxes, smart hydrological particles, air quality monitors, and intelligent well deformation sensors. These modules are designed to conduct real-time monitoring of diverse environmental parameters such as microclimate conditions, hydrological metrics, concentrations of hazardous gases, ambient air quality, and structural deformation indicators. To enhance anomaly detection accuracy and spatial resolution, the system incorporates GNSS positioning, triaxial accelerometry, and computer vision-based image recognition techniques. In addition, the system integrates large language models (LLMs) with popular communication platforms such as LINE and Messenger, enabling automated dissemination, real-time interaction, and personalized delivery of disaster-related information. This fusion significantly strengthens community-level risk awareness and emergency response capabilities. Field deployments and multi-site validations demonstrate the system’s high scalability, responsiveness, and operational robustness, underscoring its substantial potential for regional early warning applications and resilient urban planning. The findings suggest that such integrative systems can serve as a foundational infrastructure for future smart cities and disaster-resilient communities.