Intelligent Construction Schedule Dynamic Prediction System with Multimodal Data Fusion

Abstract

The construction industry faces persistent challenges in dynamic schedule management due to heterogeneous data, environmental uncertainty, and reliance on static methods, resulting in inefficiencies and schedule delays. This study addresses these gaps and proposes a multimodal data fusion (MDF)-driven framework to enhance real-time schedule prediction and decision-making. The methodology includes collecting data through APIs, IoT sensors, BIM, and RFID; pre-processing for cleansing, normalization, and integration; and training Random Forest Models to capture non-linear relationships and feature interactions to construct spatio-temporally consistent datasets. The results show that weather extremes (e.g., prolonged rainfall) have a significant impact on the foundation phase, while labor shortages and RFI upgrades are strongly associated with delays in the structural phase. A prototype system integrating BIM and real-time data visualization demonstrated its practical efficacy by providing coded progress alerts for dynamic resource optimization. This research emphasizes the transformative potential of MDF to reduce construction risk and advance intelligent schedule management. Future work should incorporate advanced AI techniques and empirical data to further improve forecasting accuracy and facilitate the development of resilient and adaptive project management systems.