Deep Learning-Enabled BIM Framework for Synergistic Carbon-Energy Optimization in Prefabricated Construction

Abstract

The aim of this study is to address the critical sustainability challenges in prefabricated construction (PC) by developing an AI-driven Building Information Modeling (BIM) system that enables collaborative prediction and optimization of carbon emissions and energy efficiency. By integrating deep learning techniques with BIM technology and PC manufacturing processes, this research examines the synergistic optimization of environmental performance across the entire lifecycle of prefabricated buildings. A comprehensive simulation test was performed on a selected residential PC project, focusing on carbon footprint reduction and energy consumption optimization. The findings revealed that the AI-driven system achieved carbon emission reductions of 17.8% and energy efficiency improvements of 22.3% compared to conventional approaches. The combination of BIM technology and deep learning models substantially enhanced sustainability performance through intelligent optimization of PC component design, manufacturing processes, and assembly sequences. Data analysis showed that the convolutional neural network (CNN) algorithm achieved prediction accuracy above 92% for both carbon emissions and energy efficiency metrics, with mean absolute percentage error (MAPE) approximately 8%. This integrated framework provides a transformative approach to achieving carbon neutrality goals in the construction industry.