Dynamic Bayesian Network Method for Structural Safety Evaluation of Diversion Tunnel

The safety evaluation of diversion tunnel structure is the basic process of diversion tunnel operation. However, due to many influencing factors and imperfect engineering data, most of the present comprehensive evaluation methods rely on the judgment of expert. This paper proposes a safety evaluation system of diversion tunnel structure, which bases on dynamic Bayesian network. Firstly, referring to several international safety standards, a three-level index system is proposed, including 7 aspects and 23 specific indexes, such as crack length and width, PH value and so on. Secondly, the comprehensive evaluation model of diversion tunnel structure safety is established. The relative importance of each index is judged by the experts, and according to these judgment data, the weight of each index is calculated by the improved fuzzy analytic hierarchy Process (FAHP). The subjective evaluation model is used to provide prior probability, and then the evaluation system based on dynamic Bayesian network(DBN) is established. The newly developed evaluation system can be updated according to the future engineering detection and evaluation data. The model is applied to the structural safety evaluation of the diversion tunnel of Houziyan Hydropower Station in China. The evaluation results show that the overall risk of the diversion tunnel is very low, but the problems such as material degradation, liner crack and liner spalling need to be solved. The evaluation result of the model is consistent with the engineering judgment, which indicates that the model has good accuracy and practicability. This study provides a possible way to evaluate engineering safety in the absence of measured data.