基于组合输入ES-GA-BP的中欧班列货运量预测

1.中南大学交通运输工程学院,湖南长沙410075;2.中南大学轨道交通大数据湖南省重点实验室,湖南长沙410075

铁路运输管理;中欧班列;指数平滑-遗传算法-反向传播神经网络;组合预测;货运量;一带一路

Freight volume forecast of China Railway Express based on ES-GA-BP with combined input
ZHANG Yinggui1,2,YANG Huiyu1,2,LEI Dingyou1,2

1.School of Traffic&Transportation Engineering, Central South University, Changsha 410075, Hunan Province, P. R. China;2.Rail Data Research and Application Key Laboratory of Hunan Province, Central South University, Changsha 410075, Hunan Province, P. R. China

railway transportation management; China Railway Express; exponential smoothing and genetic algorithm-optimized back propagation (ES-GA-BP) neural network; combined forecasting; freight volume; the Belt and Road

DOI: 10.3724/SP.J.1249.2022.02168

备注

中欧班列是推动“一带一路”倡议的重要支撑,科学合理地预测班列需求,对中欧班列运输方案的制定具有重要意义.以中欧班列铁路货运量预测为对象,统筹考虑中欧班列货运量的波动性和影响因素,提出一种基于组合输入指数平滑-遗传算法-反向传播(ES-GA-BP)神经网络的货运量预测方法.分析中欧班列货运现状,选取相关性高的影响因素作为神经网络输入;采用指数平滑法对波动较大的中欧班列货运量历史数据进行单项拟合预测,以优化神经网络输入;利用遗传算法优化反向传播神经网络参数,进一步提高预测精度;以“湘欧快线”国际运输通道货运集装箱数量预测为实例,验证方法的有效性.结果表明,组合输入ES-GA-BP方法适于解决波动较大的货运量预测问题,预测精度较好,有助于制定合理的中欧班列运输方案.
China Railway Express is an important support that promotes the Belt and Road Initiative. Scientific and reasonable forecast of the train demand is of great significance to the formulation of China Railway Express transportation scheme. Taking the forecast of freight volume of China Railway Express as an object, considering the fluctuation and influence factors, we propose a freight volume forecasting method based on exponential smoothing, genetic algorithm and optimized back propagation (ES-GA-BP) neural network with combined input. Firstly, we analyze the current status of China Railway Express freight transport, and select influencing factors with high correlation as the input of neural network. Then, the exponential smoothing method is used to fit and forecast the historical data of China Railway Express freight volume, so as to optimize the input of neural network. Genetic algorithm is used to optimize the parameters of back propagation neural network to further improve the prediction accuracy. Finally, the prediction of freight container numbers in the international transport channel of Hunan Europe Expressis taken as an example to verify the effectiveness of the method. The calculation results show that the combined input ES-GA-BP method is suitable for solving the problem of freight volume forecasting with large fluctuation, and the prediction accuracy is good, which is conducive to the formulation of a reasonable China Railway Express transportation scheme.
·