摘要
基于物流需求的实时性和不确定性,提出融合时间序列自回归-滑动平均模型ARMA和BP神经网络,构建了物流需求预测ARMA-BP模型,提出预测货物运输的ARMA-BP结合预测算法。以唐山市近几年物流运输数据为研究对象,分别运用ARMA模型、BP模型和ARMA-BP模型对物流数据进行预测分析,结果表明,与传统预测模型相比,ARMA-BP模型预测精度更高,具有一定的实用价值。
Based on the real time and uncertainty of logistics demand,the fusion time series autoregressivemoving average model ARMA and BP Neural network were proposed,and the logistics demand forecasting ARMA-BP model was modelled,An ARMA-BP combined prediction algorithm for predicting freight transport was proposed.Based on the logistics transportation data of Tangshan in recent years,using ARMA model,BP model and ARMA-BP model,the logistics data were predicted.The results show that the ARMA-BP model has higher precision and practical value than the traditional prediction model.
作者
刘凤春
赵亚宁
董新雁
刘源铄
乔鹏
谢志远
王立亚
张春英
LIU Feng-chun 1, ZHAO Ya-ning 2, DONG Xin-yan 2, LIU Yuan-shuo 2, QIAO Peng 2, XIE Zhi-yuan 2, WANG Li-ya 2, ZHANG Chun-ying 2(1.Qian College, North China University of Science and Technology, Tangshan Hebei 063000, China; 2.College of Sciences, North China University of Science and Technology, Tangshan Hebei 063210, Chin)
出处
《华北理工大学学报(自然科学版)》
CAS
2018年第3期120-128,共9页
Journal of North China University of Science and Technology:Natural Science Edition
基金
河北省自然科学基金资助(F2016209344
F2018209374)
关键词
时间序列
神经网络
残差序列
物流需求预测
time series
neural network
residual error sequence
logistics demand forecasting