基于深度学习的生鲜产品冷链物流需求预测  

Cold Chain Logistics Demand Forecasting for Fresh Products Based on Deep Learning

作　　者：郭婉萤 李滢棠[1] 白凯聪 童云河 GUO Wan-ying;LI Ying-tang;BAI Kai-cong;TONG Yun-he(School of Nautical Sciences,Jimei University,Xiamen 361021,China)

机构地区：[1]集美大学航海学院,福建厦门361021

出　　处：《物流研究》2025年第2期75-83,共9页Logistics Research

基　　金：福建省教育厅面上项目(JAT190295)。

摘　　要：近年来,深度学习发展迅速,其在预测领域的研究引起广泛关注。深度学习以其出色的泛化能力而著称,其半监督学习模式能够充分挖掘具有多维特征数据的潜力,在处理大规模数据时能展现其卓越性能。本文以生鲜产品为研究对象,探讨其物流需求预测问题。通过比较灰色预测模型、BP神经网络模型以及CNN-LSTM模型三种模型的误差,得出:CNN-LSTM模型优于灰色预测模型和BP神经网络模型,在物流需求预测方面具有显著优势,可为生鲜产品物流的规划提供可靠依据。In recent years,deep learning has developed rapidly,and its research in the field of prediction has attracted wide attention.Deep learning is known for its excellent generalization ability,and its semi-supervised learning model can fully tap the potential of multidimensional feature data,showing excellent performance when dealing with large-scale data.This paper takes fresh products as the research object to discuss the logistics demand forecasting problem.By comparing the errors of grey prediction model,BP neural network model and CNN-LSTM model,it is concluded that CNN-LSTM model is superior to grey prediction model and BP neural network model,and has significant advantages in logistics demand prediction,which can provide a reliable basis for the planning of fresh product logistics.

关 键 词：深度学习 CNN-LSTM 需求预测 灰色预测 BP神经网络 

分 类 号：F250[经济管理—国民经济]