Multi-agent based safety computational experiment system for shield tunneling projects

Hui Lu, Junxiong Qi, Jue Li, Yong Xie, Gangyan Xu, Hongwei Wang

Research output: Journal article publicationJournal articleAcademic researchpeer-review

1 Citation (Scopus)

Abstract

Purpose: In shield tunneling projects, human, shield machine and underground environment are tightly coupled and interacted. Accidents often occur under dysfunctional interactions among them. Therefore, this paper aims to develop a multi-agent based safety computational experiment system (SCES) and use it to identify the main influential factors of various aspects of human, shield machine and underground environment. Design/methodology/approach: The methods mainly comprised computational experiments and multi-agent technologies. First, a safety model with human-machine-environment interaction consideration is developed through the multi-agent technologies. On this basis, SCES is implemented. Then computational experiments are designed and performed on SCES for analyzing safety performance and identifying the main influential factors. Findings: The main influential factors of two common accidents are identified. For surface settlement, the main influential factors are ranked as experience, soil density, soil cohesion, screw conveyor speed and thrust force in descending order of influence levels; for mud cake on cutter, they are ranked as soil cohesion, experience, cutter speed and screw conveyor speed. These results are consistent with intuition and previous studies and demonstrate the applicability of SCES. Practical implications: The proposed SCES provides comprehensive risk factor identification for shield tunneling projects and also insights to support informed decisions for safety management. Originality/value: A safety model with human-machine-environment interaction consideration is developed and computational experiments are used to analyze the safety performance. The novel method and model could contribute to system-based safety research and promote systematic understanding of the safety performance of shield tunneling projects.

Original languageEnglish
Pages (from-to)1963-1991
Number of pages29
JournalEngineering, Construction and Architectural Management
Volume27
Issue number8
DOIs
Publication statusPublished - 21 Sep 2020
Externally publishedYes

Keywords

  • Computational experiments
  • Multi-agent technology
  • Safety performance
  • Shield tunneling project

ASJC Scopus subject areas

  • Architecture
  • Civil and Structural Engineering
  • Building and Construction
  • Business, Management and Accounting(all)

Cite this