Design of BIM and IoT-Based Tunnel Construction Monitoring Data Fusion and Safety Early Warning System

Abstract

The tunnel construction environment is complex, and traditional monitoring methods suffer from poor timeliness and insufficient data utilization, making it difficult to support real-time safety management and control during tunnel construction. This paper designs a construction monitoring and early warning system that integrates Building Information Modeling (BIM) and Internet of Things (IoT) technologies, adopting a “cloud-edge-device” three-tier distributed architecture. At the device layer, a multi-source heterogeneous sensor network is deployed; at the edge layer, edge computing gateways are installed for local data preprocessing and real-time analysis; at the platform layer, a data fusion analysis engine and a BIM visualization engine are integrated. For data fusion, the belief Hellinger distance is introduced to improve the Dempster-Shafer (D-S) evidence theory, effectively resolving conflicts among highly contradictory evidence sources. For safety early warning, a “4+4+N” hierarchical index system is established, and a progressively deepened three-level early warning mechanism is constructed, encompassing single-index threshold judgment, multi-parameter fusion evaluation, and Long Short-Term Memory (LSTM) time-series prediction. Finally, the system is validated through field deployment at the Yingeling extra-long tunnel in Hainan Province. The results that the proposed data fusion method achieves an accuracy of 92.5%, a 20.2% improvement over traditional D-S evidence theory. The early warning system attains an accuracy of 92.5% with a false alarm rate of 8.3%, a missed alarm rate of 3.1%, and an average response time of 2.8 seconds providing reliable technical support for tunnel construction safety management.