Intelligent code for automated code compliance checking in construction

Construction projects must comply with a multitude of building codes and standards [e.g., International Building Code (IBC), International Energy Conservation Code (IECC), and Americans with Disabilities Act Standards for Accessible Design (ADA Standards)]. Manually checking the compliance of a project with all applicable building codes and standards is costly, time-consuming, and error-prone. To reduce the time, cost, and errors of building code compliance checking, this thesis proposes a novel, intelligent-code-driven automated compliance checking (ACC) approach. First, the computable intelligent code is defined and modeled. Second, a set of computational models and methods to automatically generate the computable intelligent code are developed, including building-code requirement computability analysis, information extraction and transformation from building-code sentences, intelligent code generation, and building information modeling and regulatory information alignment. The research integrates semantic-based approaches with machine learning to automatically learn the syntactic and semantic patterns in large-scale data to improve the scalability of the proposed computational models and methods while still achieving performance levels competitive to the state of the art. To achieve the aforementioned goals, the research methodology included seven primary research tasks. First, conducting a comprehensive literature review on related topics such as ACC systems and methods and computational models and methods for text clustering, information extraction, data-to-text generation, and information alignment. Second, using a clustering-based approach to identify the different types of building-code requirement sentences in terms of their syntactic and semantic structures and computability. Third, developing a machine learning-based method to extract the regulatory semantic information elements from the textual building codes. Fourth, developing a machine learning-based method to extract the requirement hierarchy and transform the extracted information into the intelligent code. Fifth, developing a machine learning- and semantic rule-based method to align the industry foundation classes (IFC) and regulatory concepts. Sixth, developing a machine learning- and semantic rule-based generation method to generate the intelligent code from provided regulatory semantic information. And seventh, integrating the developed computational models and methods into an intelligent code-driven ACC prototype system and using the prototype in testing the performance of the integrated models and methods in noncompliance detection. The developed computational models and methods for intelligent code were individually tested, showing good performance in regulatory information extraction and transformation, IFC-regulation information alignment, and intelligent code generation, achieving over 85.0% in terms of precision- and recall-based metrics. The final integrated intelligent code-driven ACC prototype system was evaluated in checking the compliance of a BIM-based design with requirements selected from three building codes and standards (i.e., IBC, IECC, and ADA Standards) and having different levels of computability. The prototype showed good performance, achieving an average precision of 90.4%, recall of 86.3%, and F1 measure of 88.3% for noncompliance detection. Overall, the experimental results demonstrated the promise of the proposed intelligent-code-driven ACC approach.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Civil Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2022

Author dc:creator

• Zhang, Ruichuan

Contributors dc:contributor

• El-Gohary, Nora
• Ji, Heng
• El-Rayes, Khaled
• Liu, Liang Y.
• Golparvar-Fard, Mani

• automated compliance checking
• building information modeling
• machine learning
• natural language processing
• information extraction
• text clustering
• natural language generation
• information alignment

Statement dc:rights

• Copyright 2022 Ruichuan Zhang

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/115745

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F115745