Intelligent Data Framework for Crane Safety: Learning from Past Incidents to Predict Future Risks

Cranes are essential equipment in construction sites lifting and moving heavy loads. Due to the large structure, accidents can lead to significant casualties and property damage. Studies shows crane accidents are due to human error and miscommunication, and identifying the person responsible is important for preventing future accidents. A large number of reports are generated daily at construction sites, but there is no easy access interactive database available. To fill this gap, this study proposes an AI-based framework for crane accident analysis and safety management support. Using case-based reasoning (CBR) to reference past accident data, and provides interactive solutions using a large language model (LLM) which retrieves relevant information from OSHA or company guidelines through retrieval-augmented generation (RAG). Decision makers can quickly act on the proposed solutions, which can also be stored in the database to build a cyclical database to enhance crane safety and contribute to proactive accident prevention on construction sites. By pioneering new methods for analyzing textual incident descriptions tailored to safety analytics, this project introduces an innovative approach that Canadian and Korean researchers and industries can leverage to improve predictive safety measures.

Faculty Supervisor:

Jong Won Ma

Student:

Partner:

Yonsei University

Discipline:

Engineering

Sector:

Education

University:

Concordia University

Program: