Classification and application of deep learning in construction engineering and management – A systematic literature review and future innovations

Abstract

In the ever-evolving landscape of construction engineering and management (CEM), the dynamic and unique characteristics of construction project environments constantly present multifaceted challenges. These challenges are characterized by the extensive volume of project-specific information and intricate engineering data. Deep learning (DL), with its advanced analytical capabilities, has been emerging as a robust solution to these complexities. While the application of DL in CEM is on an upward trajectory, a systematic review of its implementation is conspicuously lacking. This paper, therefore, embarks on a scientometric and qualitative analysis of 296 DL-based studies related to CEM from 2014 to 2024 in the renowned data science repositories Scopus, Science Direct and Web of Science to explore the characteristics of journals, keywords and clusters. It is found that six research topics have fully utilized the advantages of DL in CEM in the last decade, including construction equipment management, structural health monitoring, construction site safety management, construction schedule management, worker health management and workforce assessment and intelligent design. Then, the studies under each research topic are summarized separately and a searchable taxonomy is proposed that secondarily categorizes each study according to the specific CEM task and DL method used to facilitate understanding and access. Finally, the primary obstacles encountered in DL itself and in its practical application in CEM are discussed. It further articulates five critical future research directions that are evolving in tandem with advances in CEM, multimodal construction site management, real-time structural health monitoring and prediction, project progress visualization and management, intelligent design with data sharing and the incorporating large language models (LLM) for text data analysis. The three goals of this study are providing CEM researchers and practitioners with an in-depth and nuanced understanding of DL, elucidating the diverse nature of CEM activities and the resulting benefits of applying DL, and identifying future opportunities for applying DL in CEM to inform subsequent ongoing academic inquiry and pragmatic applications.

Keywords

• Deep learning(DL)
• Construction engineering management(CEM)
• Condition monitoring
• Damage detection
• Large language models (LLM)