Abstract: Tunnel construction safety risks are affected by many factors, and deep mining for risk information is an effective way to improve the accuracy of risk evaluation results. The K-Means clustering algorithm was introduced into the comprehensive calculation of the traditional LEC method, and a tunnel construction safety risk evaluation model based on K-Means clustering was established. Firstly, the risks in the whole process of tunnel construction were identified by WBS-RBS method, and the risk event set was constructed. Then the accident likelihood, personnel exposure time and severity of accident consequences were taken as the tunnel construction safety risk evaluation indexes, and the linguistic multi-attribute decision method was used to transform the risk evaluation information into a data set, and the K-Means clustering algorithm were used to process the data set and determine the event risk level based on the clustering results. Finally, the propesed model was applied to the construction risk analysis of Wenbishan No. 1 tunnel cavern project. The results show that the risk evaluation results of the proposed model match with reality, and the analysis results can provide a reference for decision making in tunnel construction risk control.

Key words: tunnel engineering；highway tunnel；construction risk；risk evaluation；K-Means clustering algorithm

摘要： 隧道施工安全风险受到多方面因素影响，对于风险信息的深度挖掘是提高风险评价结果准确性的有效途径。将K-Means聚类算法引入到传统LEC法的综合计算中，建立了基于K-Means聚类的隧道施工安全风险评价模型。首先，通过WBS-RBS法识别隧道施工全过程中的风险，并构建风险事件集；然后，将事故发生可能性、人员暴露时间和事故后果严重程度作为隧道施工安全风险评价指标，利用语言型多属性决策法把风险评价信息转化为数据集，采用K-Means聚类算法对数据集进行处理，基于聚类结果确定事件风险等级；最后，将模型应用于文笔山1号隧道洞口工程施工风险分析中。结果表明：所建模型风险评价结果与现实相吻合，分析结果可为隧道施工风险控制提供决策参考。

关键词: 隧道工程；公路隧道；施工风险；风险评价；K-Means聚类算法