基于Xgboost的铁路货运装卸作业时间预测  

Loading and Unloading Time Estimation Based on Xgboost for Railway Freight Transport

作　　者：周瑾 王文斌 刘子扬 刘永壮 Zhou Jin;Wang Wenbin;Liu Ziyang;Liu Yongzhuang(CRSC Research&Design Institute Group Co.,Ltd.,Beijing 100070,China;China Shenhua Energy Co.,Ltd.,Beijing 100040,China)

机构地区：[1]北京全路通信信号研究设计院集团有限公司,北京100070 [2]中国神华能源股份有限公司,北京100040

出　　处：《铁路通信信号工程技术》2024年第3期36-41,共6页Railway Signalling & Communication Engineering

基　　金：中国神华重载铁路运输大数据分析平台研究项目(SHGF-22-02)。

摘　　要：传统方法推算货运装卸作业时间直接使用站细里的标准时间,难以刻画复杂因素影响下作业时间的变化情况,准确率较低。通过数据挖掘的方法,收集铁路综调信息系统记录的货运装卸作业相关数据,利用增强决策树模型Xgboost学习装卸作业相关影响因素对其作业时间的影响,实现货运装卸作业时间预测,对比基线模型准确率提升明显,能更有效辅助车流推算与运行图自动编制。The traditional method of loading and unloading time prediction of freight transport,which directly utilizes the standard time specifi ed in the Detailed Instructions Governing Train Operation at Station,cannot properly characterize the time change under the impacts of complex factors,and achieves low prediction accuracy.This paper utilizes the data mining method to gather the relevant data on the loading and unloading time of freight transport from the railway integrated dispatching information system.It also utilizes the boosted decision tree model Xgboost to predict the loading and unloading time of freight transport.Compared with the reference model,the proposed model can achieve substantial improvement in prediction accuracy,and provide more eff ective support for traffi c fl ow prediction and automatic drawing of train operation charts.

关 键 词：重载铁路 货运作业 装卸作业时间 决策树 Xgboost 

分 类 号：U294.27[交通运输工程—交通运输规划与管理]