A dual-loop collaborative framework for smart construction sites

This study addresses the challenges of real-time response and closed-loop decision-making in smart construction. We propose a dual-loop collaborative framework integrating Building Information Modeling (BIM), the Internet of Things (IoT), and LSTM-Attention networks, validated via a 261-sensor deployment in the Taipei Bridge and Tunnel Joint Project. The architecture achieves sub-second synchronization and enables multi-objective optimization using NSGA-II. Experimental results demonstrate statistically significant improvements (p < 0.001): a 46.2% reduction in system latency, a 41.4% decrease in safety incidents, an 18.5% reduction in direct costs, and an 11.1% decline in CO₂ emissions, while maintaining a 99.1% safety compliance rate. The framework successfully bridges the gap between digital models and physical realization, providing a scalable solution for adaptive and sustainable construction management.