基于SVM的城市地下工程施工安全风险预测研究

SVM-based Method for Predicting Safety Risk of Urban Underground Engineering Construction

为科学准确预测城市地下工程施工安全风险水平,开发一种基于支持向量机(SVM)模型的预测方法。利用Citespace对近5年相关文献进行了主题和关键词聚类分析,将聚类分析结果作为参考。基于4M1E事故要素理论,结合城市地下工程特点和专家意见,建立了契合地下工程施工实际的安全风险评价指标体系,从而构建了安全风险水平预测SVM模型。以武汉、北京、广州等地区范围内的25个城市地下工程施工项目统计数据对SVM模型进行学习训练和测试验证。结果表明:SVM模型训练集回代检验和测试集预测结果均与实际情况完全一致,可作为预测城市地下工程施工安全风险水平的有效方法。

In order to scientifically and accurately predict the safety risk level of urban underground engineering construction,a prediction method based on the support vector machine(SVM)model was developed.A cluster analysis of themes and keywords was carried out on related papers in the past 5 years through Citespace,and use the results of cluster analysis as a reference.Based on the 4M1E accident element theory,and combined with the characteristics of urban underground engineering and expert opinions,a safety risk evaluation index system that fits the actual construction of underground engineering was established,thus constructed an SVM model for predicting safety risk levels.The SVM model was trained and tested with statistical data of urban underground construction projects in 25 cities in Wuhan,Beijing,Guangzhou and other regions.The results showed that the results of the back test of training set and prediction of test set of the SVM model were completely consistent with the actual situation.Thus,SVM can be used as an effective method to predict the safety risk level of urban underground engineering construction.