BIM데이터 품질검토 자동화를 위한 정형 문법 기반 검사 규칙 프레임워크 연구

Formal Grammar-Based Inspection Rule Framework for Automated BIM Data Quality Management

This study proposes a format grammar-based quality inspection framework to overcome the inefficiencies caused by the heterogeneity and tool dependency of various BIM data formats, such as IFC and COBie. Existing quality inspection methods rely on format-specific scripts, which prevent the creation of an independent, consistent rule set for quality management. To solve this, the research introduces a structured, key-value-based rule grammar that functions independently of the BIM data model. The framework supports multiple condition types (e.g., range, list, equation) and uniquely integrates a Large Language Model (LLM) to interpret rules requiring contextual or semantic judgment. This allows for the automation of complex, text-based regulation checks that are difficult for traditional rule-based systems. A rule interpretation engine processes these rules, and users can upload BIM files, visually review the inspection results, and export reports in PDF, Excel, or BCF formats. Experimental results confirm that a single ruleset can be applied effectively across different BIM formats. The LLM demonstrated high potential for automating semantic-based inspections, proving more effective than existing methods. This framework serves as a foundational technology for systematically and reusably managing the quality of diverse BIM models throughout the design, construction, and maintenance phases.