Machine learning model for the BIM classification in IFC format

Abstract

In the rapid development of information technology in the field of Building Information Modeling (BIM) there is a growing need for efficient classification of construction information. One of the key steps to move towards digital construction involves creating reliable systems for classifying BIM elements, providing the foundation for various use cases, from facilitating model navigation to obtaining practical outcomes such as cost estimates and materials quantities. However, the BIM classification process in practice is labor-intensive and time-consuming and leads to an increase in the cost. This study explores the application of an innovative method, based on artificial intelligence algorithms. This method automates the assignment of codes to information model components. The research investigates classification systems, machine learning models and selects the most accurate one for the classification task. It is based on metrics such as accuracy and F1-score in order to achieve an optimal balance between the efficiency and accuracy according to predefined parameters. The article presents software for automatic prediction and assignment of codes in accordance with the selected classifier, developed on selected algorithms.

Keywords

• classification in construction
• artificial intelligence
• classifier
• classification model
• machine learning
• neural networks
• bim technologies
• bim
• ifc
• revit
• civil engineering
• сad