AI-based BIM model quality verification for construction industry

Abstract

Building Information Modeling (BIM) has emerged as a powerful technology for enhancing collaboration and efficiency in the architecture, engineering, and construction industry. As BIM grows in complexity, ensuring their quality becomes critical to avoid errors and inconsistencies that may impact project outcomes. This paper presents an integrated new BIM model quality definition theory and on top of it, a practical AI-based approaches for BIM quality checking methods, which generates significant benefits associated with this new BIM quality evaluation approach. This research explores the major issues in BIM quality and its impact on current industry practices while presenting a brand-new BIM quality definition - the EIT theory which covers BIM model object elements(E), information(I) and topology(T). In addition, exploring the application of Artificial Intelligence (AI) techniques, especially in natural language processing (NLP) to automate and augment the quality checking process in BIM models. By leveraging BIM modeling processes and regulations, natural language processing algorithms convert BM rules and identify potential quality issues, such as geometric errors, attribute inconsistencies, clashes, and violations of industry standards. Following the new theory, a comprehensive AI based BIM model checking process is developed and applied to a real estate organization as a pilot study, including enhanced BIM standards, BIM guides, evaluation software development and quality insurance collaboration processes. The result of this case study further proves the advantage of this EIT theory and the potential gains in efficiency, accuracy, and scalability that AI brings to the quality checking process. After the development of EIT based AI BIM quality check process, this paper also discusses the challenges associated with AI-based quality checking, such as the requirement of BIM modeling standards, the need for high-quality training data, the interpretability of AI outputs, and the roles of domain expertise in result validation, highlights the interdisciplinary collaboration to integrate AI into existing BIM workflows. The findings of this research contribute to the existing body of knowledge on AI applications in the construction industry, especially the BIM model quality definition and quality checking. The utilization of these AI techniques significantly enhances the effectiveness and reliability of real project BIM quality assessment processes to improve overall project outcomes with BIM. This research concludes by identifying future research directions, including exploring advanced AI algorithms, integrating AI with BIM tools, and developing standardized evaluation metrics for quality checking methodologies, further advancing the field and facilitating the widespread adoption of AI with BIM practices.