Safety risk assessment of shield tunneling under existing tunnels: A hybrid trapezoidal cloud model and Bayesian network approach

To ensure the construction of new tunnels, it is very important to assess the risk of shield tunneling under existing tunnels. A hybrid approach that integrates a trapezoidal cloud model (TCM) and a Bayesian network (BN) is proposed for the safety assessment of shield tunnels crossing under existing tunnels. According to engineering practice and literature review, 12 risk factors are adopted as evaluation indexes, and the TCM is used to transform these factors, discretize the continuous nodes, and improve the accuracy of the prior probability. Forward reasoning, sensitivity analysis, intensity analysis and reverse diagnosis are carried out for the event via the TCM-BN, and corresponding risk control measures are taken to realize the comprehensive evaluation and control of the dynamic risk of constructing an underpass. Taking the Wuhan Rail Transit project as an example, the results indicate that (1) the safety risk level of the monitoring points obtained through the prior probability prediction with the TCM-BN model is more accurate than the on-site assessment results. (2) Risk diagnosis based on the TCM-BN model can determine the most unfavorable factors, and the four key sensitive points are identified through sensitivity analysis. (3) Based on the analysis results, targeted risk control measures and real-time risk assessment are realized, and the proposed method can provide a new tool for the safety risk assessment of tunnel construction under existing tunnels.