Automated BIM Object Classification to Support BIM Interoperability

The ISO-registered industry foundation classes (IFC) data standard can facilitate building information modeling (BIM) interoperability by allowing a “one-to-many” software information flow pattern in the architecture, engineering, and construction (AEC) industry instead of the more complex “many-to-many” information flow pattern. However, even a full adoption of IFC cannot guarantee a seamless BIM interoperability, because of the flexibility allowed in adopting/using the IFC standard. Object classification in BIM is an important part of BIM interoperability. As part of an initial effort to pursue seamless BIM interoperability, the authors propose to use intrinsic geometric representations of IFC objects and geometric theorems to support the automated BIM object classification. A six-step method was proposed to develop automated IFC object classification algorithms using a data-driven approach. The method was preliminarily tested on classifying IFC objects into a cone frustum shape. The developed algorithm was successfully tested on three unseen bridge data. This shows an early promise of the proposed method to support error-free BIM object classification.