Risk Analysis For High-speed Rail Karst Tunnel Based On The Bayesian Network

In the southwest of China, the geological condition is complex, covering withmuch karst which increases the unpredictability of risk. Traditional methods of riskanalysis primarily rely on the subjective experience of the expert which has largelimitations. This paper introduces the Bayesian network theory and establishes theBayesian network model to investigate and analysis the karst tunnel surroundingrock classification and surrounding rock stability which also obtain the inspection ofTSP system and monitoring measurement.Yun Gui high-speed rail tunnel construction—luo de yi tunnel has been takenas the background of this paper.Aiming at analyzing for the stability of karst tunnelsurrounding rock classification and surrounding rock,and the whole paper is dividedinto four parts as followed:1. This paper reviews the current studies on risk management in tunnelingapplication and the theory research and application of Bayesian network at home andabroad detailedly.2. This paper discusses the basic theory of risk analysis and Bayesian networkin detail and illuminates the contents, steps and methods of risk analysis. Beside,Comparative analysis showes that merits of the Bayesian posterior probabilitymethod. And pay more attention on the concept of Bayesian network theory,Bayesian networks model and its probability reasoning and simplification. Moreover,it also introduces the Bayesian network model software Netica in detail.3. The paper discusses comprehensive evaluation methods for surroundingrockmass classification.The methods discover the surrounding rock classificationindicators and their relation. And finally the Bayesian network of surrounding rockclassification in karst tunnel is established. Using three functions of Netica such asposterior probability reasoning, most probable explanation and sensitivity analysis,we can deduce a conclusion that the karst is the main factor to affect tunnelsurrounding rock classification.In addition, the TSP proves the Bayesian networkmodel in the prediction of the surrounding rock classification.4. This paper investigates natural factors of engineering geology and humanfactors of engineering activities to ascertain the index which will influence karsttunnel surrounding rock stability. Combined with the construction situation,wemainly select indexs as the node such as the arching, karst development, excavation section, excavation disturbance, supporting strength, and supporting time; Using theBayesian network model of the risk analysis of karst tunnel and basing on theⅣ-class surrounding rock stability of karst tunnel as the research object to analysisnatural and human factors which affect the stability of karst tunnel surrounding rock.And the result shows that the karst has a critical impact on stability of surroundingrocks. And the TSP proves the applicability to dealing with the risk of Bayesiannetwork model.