Advancing Logistics Management: E3L-Net for Predictive Demand Analytics

The essence of smart logistics lies in leveraging information resources and intellectual assets to efficiently and precisely match the multidimensional demands and supplies within the logistics system. Unlike supply, demand is more dynamic, making the accurate capture and prediction of demand variat...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access 2024, Vol.12, p.114809-114819
Main Author: Lu, Yao
Format: Article
Language:English
Subjects:
Advancing logistics management
Analytical models
Data models
Deep learning
Feature extraction
Logistics
Long short term memory
LSTM
Mathematical models
Predictive models
smart logistics
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The essence of smart logistics lies in leveraging information resources and intellectual assets to efficiently and precisely match the multidimensional demands and supplies within the logistics system. Unlike supply, demand is more dynamic, making the accurate capture and prediction of demand variations across different levels and time dimensions the core and key to developing smart logistics systems. Temporary micro-level demand prediction not only enhances the timeliness, accuracy, and cost-effectiveness of micro-level supply but also reveals macro trends and extended patterns of logistics demand, providing decision support for logistics management at all levels. This study addresses the challenges in predicting temporary logistics demand, characterized by variability, high stochasticity, and abrupt changes. We propose an advanced E3L-Net model, combining ensemble empirical mode decomposition, local mean decomposition, long short term memory networks, and local error correction. The E2L-Net model, formed by integrating ensemble empirical mode decomposition and local mean decomposition, decomposes the original data to stabilize it and mitigate endpoint effects. LSTM is then used to predict these decomposed signals, leveraging its superior temporal modeling capabilities. The LEC model further refines these predictions by correcting local abrupt changes. Our experimental analysis, utilizing logistics demand data from a company, demonstrates that the proposed model significantly outperforms 11 other models, highlighting its effectiveness and generalization capability in handling temporary logistics demand predictions.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2024.3444282