基于地铁施工事故案例的机器学习灾害预测方法

针对城市轨道交通施工事故的频繁发生,建立了基于地铁施工事故案例的机器学习灾害预测模型.通过收集过往地铁施工事故案例建立数据集,引入天气气象、水文地质、周边环境、施工因素等外部风险源做为特征,分析决策树、随机森林、SVM、XGBoost等灾害预测模型对施工事故的预测能力.结果 表明经过网格搜索后XGBoost的预测效果明...

Machine Learning Disaster Prediction Method Based on Subway Construction Accident Cases

Aiming at the frequent occurrence of urban rail transit construction accidents,a machine learning disaster prediction model based on subway construction accident cases has been established Establish a data set by collecting past subway construction accident cases,introducing weather and meteorology,hydrogeology,surrounding environment,construction factors and other external risk sources as features,and analyzing the prediction of construction accidents by disaster prediction models such as decision trees,random forests,SVM,XGBoost,etc.ability.The results show that the prediction effect of XGBoost after grid search is significantly higher than other models.Its macro-average AUC value and micro-average AUC value reach 0.7564 and 0.8624 respectively,which are 0.87% and 12.99%higher than the second-ranked model,respectively.