摘要
新冠疫情期间,公路货运量明显下滑,公路运营状况变化复杂,亟须科学预测公路货运量。通过灰色关联分析,确定疫情期间公路货运量主要影响因素,构建了基于灰色组合(GC)-修正BP神经网络(rBPNN)模型的公路货运量预测方法。以我国2017年7月—2020年5月的公路货运量统计数据为原始数据,对BP神经网络进行训练和检验,并引入修正系数H_(M)对预测结果进行修正。以疫情期间近5个月数据为基础,用灰色组合模型预测下月公路货运量各主要影响因素值,再运用修正BP神经网络预测我国2020年6月的公路货运量。将GC-rBPNN模型与其他预测方法进行对比分析,GC-rBPNN模型的PE和MAPE分别为0.21%和3.21%,结果表明,GC-rBPNN模型的预测精度更高,有一定的可行性和有效性。
Under the COVID-19 epidemic situation,the volume of road freight has declined significantly,and road operations have changed complexly.It is urgent to scientifically predict the volume of road freight.Through gray correlation analysis,the main factors affecting road freight volume during the epidemic period are determined,and a road freight volume forecast method based on the gray combination(GC)-revised BP neural network(rBPNN)model is constructed.The BP neural network is trained and tested based on the statistical data of China’s road freight volume from July 2017 to May 2020 as the original data,and the“correction coefficient”H_(M) is introduced to modify the predicting result.Based on the data of the past five months during the epidemic,the gray combined model is used to predict the value of the main factors affecting the road freight volume in the next month,and the BP neural network is used to predict China’s road freight volume in June 2020.Compared the GC-rBPNN model with other prediction methods,the PE and MAPE of the GC-rBPNN model are 0.21%and 3.21%,respectively.The results show that the prediction accuracy of the GC-rBPNN model is higher,and the method has certain feasibility and effectiveness.
作者
田晟
李成伟
黄伟
王蕾
TIAN Sheng;LI Chengwei;HUANG Wei;WANG Lei(School of Civil Engineering and Transportation,South China University of Technology,Guangzhou Guangdong 510640,China;Guangzhou Transportation Design&Research Institute Co.,Ltd.,Guangzhou Guangdong 511430,China)
出处
《广西师范大学学报（自然科学版）》
CAS
北大核心
2021年第6期24-32,共9页
Journal of Guangxi Normal University:Natural Science Edition
基金
国家留学基金(20170615503)
广东省科技计划(2015A080803001)
广东省自然科学基金(2020A1515010382)。
关键词
公路货运量
疫情
灰色关联度
BP神经网络
组合预测模型
highway freight volume
epidemic situation
grey correlation degree
BP neural network
combined forecasting model
