The Application Of Bayesian Network Based On Fault Tree Analysis Method In Tunnel Construction Risk Management

Since the 21 st century,with the leap forward development of basic traffic engineering construction in our country,the urban underground pipe and culvert tunnel?railway tunnel?highway mountain tunnel and high difficulty tunnel of across rivers and lakes,the engineering construction has entered a stage of rapid development.And this undoubtedly proposed higher requirements of tunnel construction quality,safety and driving comfort and aesthetic and some other aspects.Because of China's vast territory,mountains,complex terrain conditions,tunnel construction inevitably need to across the mountains.Due to the highway mountain tunnel in addition to its main engineering geological problems,such as under the condition of high ground stress prone to rock burst or large deformation,joint fissure development easily lead to deformation of surrounding rock collapse,instability,including by the special geological disaster,such as coal seam gas escaping the explosion,goaf collapse instability,large-area subsidence of ground surface,water gushing and other.In recent years in the practice of our country from these sources of risk caused by the corresponding engineering accident more and more.Therefore,the safety of tunnel engineering challenges becoming higher,it is urgent need to establish a set of perfect of highway tunnel construction risk assessment system.This paper based on the construction process of Shujiawan highway tunnel from Guiyang to Qianxi,established a set of tunnel construction risk management system,applied to the risk management process of this tunnel,the main research results are as follows:(1)This paper use a recognition method which combine expert investigation method and risk source census form method,according to the form of common risk event,it is concluded that the Shujiawan tunnel construction process may occur six risk factors,they are gas explosion,landslide,surface subsidence,large deformation,fire and traffic accident risk,then set up the risk events fault tree diagrams.(2)The traditional fault tree analysis method is a method of binary analysis,all basic event is only "happen" and "not" two states;Bayesian network analysis method is a kind of directed graph reasoning method based on network structure,it has a strong reasoning ability,can be integrated by conditional probability considering basic event interaction effects.This paper combined the advantage of fault tree analysis method and the bayesian network analysis method,puts forward the improved bayesian network based on fault tree analysis method.(3)Combined with fault tree diagram,established the risk events bayesian model which based on fault tree analysis method using bayesian simulation software Netica to calculate a quantitative calculation of each risk events and then get diagnostic reasoning of each event,it is concluded that,the probability of each event,risk grade and the sensitivity of the top event to occur.Based on the binary part of the event for the three states correction,get a more detailed results.(4)In allusion to the relationship of the sensitive degree between each risk event and bottom events,different risk event countermeasures are given.