交通运输系统工程与信息 ›› 2019, Vol. 19 ›› Issue (3): 111-118.

• 系统工程理论与方法 • 上一篇    下一篇

基于Box-Cox-TQR概率密度的铁路运用车保有量预测方法

李夏苗*,王丽珊,郭旺   

  1. 中南大学 交通运输工程学院,长沙 410000
  • 收稿日期:2018-10-25 修回日期:2019-02-22 出版日期:2019-06-25 发布日期:2019-06-25
  • 作者简介:李夏苗(1963-),男,湖南茶陵人,教授,博士.
  • 基金资助:

    国家自然科学基金/National Natural Science Foundation of China(U1334207).

A Method to Predict the Number of Rail Freight Serviceable Cars Held Kept Based on Box-Cox-TQR Probability Density

LI Xia-miao, WANG Li-shan, GUO Wang   

  1. School of Traffic and Transportation Engineering, Central South University, Changsha 410000, China
  • Received:2018-10-25 Revised:2019-02-22 Online:2019-06-25 Published:2019-06-25

摘要:

合理的铁路运用车保有量,对满足铁路货运需求,提高货车运用效率,降低运营成本等有重要作用.考虑铁路运输系统复杂的内外部环境及其动态变化特性,对影响运用车保有量因素定性分析;提出了粗糙集属性约简、灰色关联分析、逐步回归方法相结合的主要影响因素识别方法.以此为基础,建立了基于 Box-Cox变换分位数回归(Box-Cox-TQR)和核密度估计相结合的概率密度预测模型.以国家铁路局运用车保有量实际数据为基础,进行预测试验.结果表明,利用主要因素识别的方法符合目标值的运动变化规律,预测结果具有良好的精度.此外,概率密度预测比点预测、区间预测传递出更多信息,为管理决策提供更多准确有用信息.

关键词: 铁路运输, 运用车保有量, 概率密度预测, 主要因素识别, 变换分位数回归

Abstract:

The rational number of serviceable cars held kept plays an important role in meeting railway freight demand, improving efficiency and reducing transportation cost. Considering the complex internal and external environment of railway transportation system and its dynamic change characteristics, this paper makes a qualitative analysis of the factors affecting the number of serviceable cars held kept. A method for identifying the main influencing factors combining rough set attribute reduction, grey correlation analysis and stepwise regression is proposed. Based on this, a probability density prediction model based on Box-Cox transform quantile regression (Box-Cox-TQR)and kernel density estimation is established. Based on the actual data of the number of serviceable cars held kept of the State Railway Administration, the prediction test is carried out. The results show that the identification method using the main factors conforms to the motion variation law of the target value and have good accuracy. In addition, probability density prediction conveys more information than point prediction and interval prediction, providing more accurate and useful information for management decisions.

Key words: railway transportation, number of serviceable cars held kept, probability density prediction, main factors identification, transform quantile regression

中图分类号: