Prediction and analysis of tunnel water inrush disasters in Chinese Karst area based on Variable weight-weighted Bayesian network model

Zengguang Xu,Fanhua Kong,Cheng Cao,Zeyuan Zhang
DOI: https://doi.org/10.1007/s13146-024-01031-7
2024-12-08
Carbonates and Evaporites
Abstract:The occurrence of water inrush presents a significant hazard during karst tunnel excavation, particularly in karst water-rich areas. However, existing prediction methods lack quantitative assessment and uncertainty treatment. In this paper, a risk prediction model of tunnel water inrush in karst water-rich areas is proposed, which combines Variable weight theory, Game theory and Bayesian network. Firstly, based on the collection and analysis of 91 cases of water inrush incidents, a comprehensive water inrush assessment index system consisting of 10 indicators was established by taking into account engineering geology, hydrogeology, and human influence conditions. Secondly, the subjective and objective weight vectors were optimized using the Game theory combination assignment method, and the field-measured indicators were normalized for calculation of variable weights. Finally, the Weighted Bayesian Network model was utilized to predict the probability of water inrush occurrence and the probability within each graded range of water inrush. The model was utilized to forecast and evaluate the potential for water inrush, with specific reference to the Jigongling Tunnel and the Shanggaoshan Tunnel as case studies. The findings demonstrate that the water inrush safety assessment model, which is based on Variable weight—weighted Bayesian network model and proposed in this study for tunnel construction in karst areas, aligns well with the results of on-site excavation. This validation confirms the practicality and reliability of the prediction and assessment model, as well as enables the prediction of water inrush volume range while predicting the risk of water inrush occurrence.
geology
What problem does this paper attempt to address?