Abstract In order to evaluate the risk level of water inrush caused by karst cave accurately and effectively, a novel quantitative assessment model was established based on the reliability theory and genetic algorithm-back propagation (GA-BP) neural network. First, the reliability theory and the calculation formula of the minimum safe thickness were used to calculate the water inrush probability. Second, the GA-BP neural network was applied to predict the disaster consequence caused by water inrush. Six factors, including water pressure, hydraulic supply, type of gap, filling situation, degree of water enrichment and reserves of cave, were selected as the input layer of the neural network. The disaster consequence was selected as the output layer. Similar projects were screened to obtain statistical information for indices, and the Normand function in MATLAB was used to transform the information into quantitative data. Finally, the model was established by combining the probability and disaster consequence of water inrush. The 602cave in Yesanguan tunnel was taken as an engineering sample to verify the feasibility of the novel model. The obtained results showed that the proposed model is comprehensive and accurate in quantitative assessment, which has good application prospects in engineering.

中文翻译：

---

基于可靠性和GA-BP神经网络的隧道溶洞突水风险评估

摘要 为准确有效地评价溶洞突水风险等级，基于可靠性理论和遗传算法-反向传播（GA-BP）神经网络，建立了一种新型的定量评价模型。首先，利用可靠性理论和最小安全厚度计算公式计算突水概率。其次，应用GA-BP神经网络预测突水造成的灾害后果。选取水压、水力供给、缝隙类型、充填情况、富水程度和洞穴储量6个因素作为神经网络的输入层。选择灾难后果作为输出层。筛选类似的项目以获得指数的统计信息，并使用MATLAB中的Normand函数将信息转化为定量数据。最后结合突水发生的概率和灾害后果建立模型。以野三关隧道602洞为工程样例，验证了新模型的可行性。所得结果表明，所提出的模型在定量评价方面全面准确，在工程中具有良好的应用前景。