Integration of Building Information Modeling and Artificial Intelligence Systems to Create a Digital Twin of the Construction Site

This study is devoted to the development of an information system for creating digital twins of construction sites by integrating Building Information Modeling (BIM) technologies and artificial intelligence systems. An artificial intelligence system has been developed that combines the Convolutional Neural Network (CNN) named “You Only Look Once” v3 (YOLOv3) and Feed-Forward Neural Network (FFNN) architectures as a comprehensive mechanism for detecting, classifying, and evaluating individual components, objects, systems and processes of the construction project at all stages of its life cycle. The article describes the inclusion of the Internet of Things (IoT) and Big Data Technologies for recognizing the physical representation of the large-scale array of construction objects used on a construction site in real-time. The effectiveness of identifying the correspondence of a set of BIM-model attributes that provide the circumstances for creating a digital twin of the construction site is determined. The reliability of object recognition by the YOLOv3 model is proved, which is confirmed by the map indicator – 90.04%, which additionally correlates the FFNN model with attributes of the reference BIM model. The results of this study can be used to further improve the concept of developing digital twins using other relevant and key components of the construction project and the latest information technologies.