Safety risk evaluation of tunnel collapse based on Bayesian network of improving conditional probability

To achieve a more accurate and effective safety risk evaluation for tunnel collapse, a new method to evaluate the possibility of tunnel collapse was proposed by improving the conditional probability determination process of Bayesian networks. 5 intermediate events and 23 basic events were selected to construct the Bayesian network structure of tunnel collapse. The membership degree of the root node event belonging to each standard risk level was determined by connection clouds, and then the prior probability of the node was determined by the membership degree. Part of expert decision information with minimum complexity was fused according to DS evidence theory, and the conditional probability determination process was improved by combining the weights of node events and the "state risk contribution value". This method was applied to Laoshan tunnel project in Labin, Guangxi Province, and to calculate the risk probability of tunnel collapse and diagnose the cause of the accident. The results show that the risk level of tunnel collapse belongs to "high risk". The fault fracture zone, intensity of surrounding rock, excavation cycle footage and construction step are the main reasons for tunnel collapse. Finally, this method is compared with traditional method, showing that the method presented in this paper has higher computational efficiency and accuracy than traditional method. © 2023 Central South University of Technology. All rights reserved.