Multisource information fusion for safety risk assessment in complex projects considering dependence and uncertainty

Area of Science:

• Civil Engineering
• Risk Management
• Data Science

Background:

• Tunneling projects are vital for infrastructure, but leakage risks pose significant challenges.
• Uncertainty and fuzziness in risk factors complicate accurate assessment.

Purpose of the Study:

• To propose a hybrid approach for robust tunnel leakage risk assessment.
• To explore factor dependence and fuse multisource information effectively.

Main Methods:

• Integration of copula theory, cloud model, and risk matrix.
• Development of a risk index system with nine critical factors.
• Application of Sobol-enabled global sensitivity analysis (GSA).

Main Results:

• Identified risk statuses for three tunnel sections as Grade I (safe), II (low-risk), and III (medium-risk).
• Waterproof material degradation identified as a critical factor impacting tunnel sections.
• Confirmed strong interactions between influential factors and the necessity of studying factor dependence.

Conclusions:

• The hybrid copula-cloud model provides a robust method for tunnel risk assessment, considering factor dependence and uncertainty.
• The proposed approach enhances understanding of risk trends and factor contributions.
• The neutral risk matrix demonstrates robustness and high recognition capacity in risk assessment.