Citation

Ma G, Wu Z, Jia J, Shang S. Safety Sci. 2021; 143: e105410.

Copyright

(Copyright © 2021, Elsevier Publishing)

DOI

10.1016/j.ssci.2021.105410

PMID

unavailable

Abstract

The analysis of safety risk factors in the process of construction project management has strong subjectivity and locality, which fails to fully mine the valuable information contained in heterogeneous data. Consequently, the purpose of this paper is to develop an integrated analysis framework (IAF) to identify and evaluate potential risk factors in construction projects. In this framework, machine learning is used to mine safety risk factors from heterogeneous data, and the initial importance of risk factors is calculated based on two judgment categories. Meanwhile, the concept of cascading effect is introduced to establish an activity-on-the-node (AON) network to furtherly evaluate the negative impact of risk factors on project safety. The case study results show that machine learning can more efficiently mine the associated risk factors and effects with cascading failures, which indicates that the data-based safety risk factor analysis is more objective and accurate.

Language: en

Keywords

Cascading effect; Heterogeneous data; Machine learning; Risk factors