Abstract: With the construction of a country with a strong transportation network, a large number of tunnels have to be built and tunnel slag has to be tackled in the waste dump in the construction of an increasing number of major linear infrastructure projects such as roads and railways across the western region. Under the joint influence of natural environment and human activities, many large-scale waste dumps located at complex and dangerous environment areas pose a great threat to soil and water conservation, ecological environment, life and property, due to potential risks of collapse, landslide and debris flow. From a system perspective, a conceptual framework of waste dump system risk analysis is constructed. Then, based on the life cycle stages of waste dump system, the logical chain of constructing fault tree waste on dump system is proposed with the risk causing management behavior and natural environmental factors. After that, by setting logical relationship transformation rules, the fault tree is transformed into a Bayesian network structure, and the expert evaluation method is introduced to determine the conditional probability between Bayesian network nodes for diagnosing and predicting the system risk of mega-project waste dump. Finally, the proposed modeling work at the integration of fault tree and Bayesian network is illustrated with an example of a waste dump, showing how to support the risk management of mega-project waste dump under complex and dangerous environmental conditions.

Key words: mega-project, waste dump, risk, Bayesian network, fault tree

摘要： 伴随建设交通强国，越来越多的穿越西部地区的公路、铁路等线性基础设施类重大工程均需要建设大量隧道并通过弃渣场安置隧道洞渣。相应地，很多大规模弃渣场不可避免选址在复杂艰险环境地区，在自然环境和人类活动的共同影响下，弃渣场潜在的崩塌、滑坡和泥石流风险，对水土保持、生态环境和生命财产威胁巨大。首先，从系统视角出发，构建弃渣场系统风险分析逻辑框架；然后，根据对弃渣场系统的生命周期阶段划分，结合引起风险的管理行为和自然环境因素，提出弃渣场系统故障事件识别分析逻辑框架，用于构建弃渣场系统风险分析故障树。进一步，通过逻辑关系转化规则设定，将故障树转化为贝叶斯网络结构，并引入专家评判确定贝叶斯网络节点间的条件概率，将专家知识转化为用于诊断与预测重大工程弃渣场系统风险的贝叶斯网络模型。通过一个大型弃渣场的灾害实例进行了说明，揭示了如何通过故障树和贝叶斯网络将专家知识进行综合集成，以支撑复杂艰险环境条件下重大工程弃渣场的风险管理。

关键词: 重大工程, 弃渣场, 风险, 贝叶斯网络, 故障树