PREDICTING MODEL OF ACCIDENT RATE FOR FREEWAY LONG TUNNEL USING DYNAMIC FAULT TREE AND BAYESIAN NETWORK

Considering the driving dynamics and time sequences, an accident rate predicting model is put forward with time dimension based on Dynamic Fault Tree (DFT) and Bayesian Network (BN). The dynamic logic gate is employed to express a variety of main factors and the time logic relationship of the accidents in the freeway long tunnel. By modifying the probability of the elementary event, the DFT is transformed into the BN, and then the calculation method of probability of top event is given. With the introduction of the effective probability of the prevention and emergency as well as impact probability, the prevention and emergency measures are taken into Fault Tree (FT) to investigate the relationship between the accident probability and the measures. Taking a long tunnel of freeway in Shaanxi province for instance, it can be concluded that taking preventive measures only and taking emergency measures only can reduce accident rate by 37.45 and 48.87%, respectively. However, taking both measures can reduce accident rate by 86.83%. This study offers a useful suggestion for the improvement of tunnel management.