A BIM-based 3D design compliance review system for power grid projects

Based on the background of smart grid and information construction, this study addresses the inefficiency and inaccuracy of traditional manual compliance review in BIM-based power grid engineering, particularly stemming from the laborious and error-prone human interpretation of complex, textual design specifications. To bridge this gap, Natural Language Processing (NLP) and deep learning are employed to automatically extract, interpret, and formalize complex textual specifications into quantifiable design rules. This process builds a specification knowledge base, which is then integrated with a BIMBase-based review platform, enabling real-time compliance checking. By mapping model components to specification clauses, the system supports automatic validation and error feedback. A case study on a Xuzhou substation demonstrates the platform’s effectiveness. Quantitative results show that the automated system reviewed 720 components in 30 minutes—six times faster than manual review—and detected 80 problematic components versus 52 manually, achieving an F1 score of 89.06%. The proposed method improves the accuracy, efficiency, and standardization of compliance checks. This research provides a practical solution for intelligent BIM review in power grid engineering, contributing to safer and more reliable infrastructure development.