Article Details

Abstract

As an important part of modern urban logistics system, short-distance transportation directly affects the capacity scheduling and overall operation efficiency. Aiming at the violent fluctuation and nonlinear characteristics in the volume data, this paper designs a composite prediction framework combining AutoRegressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) neural network. By analyzing the time series characteristics of historical cargo volume data, the stationarity test is carried out and the preliminary trend model is established. The ARIMA model is used to capture the linear dynamic characteristics, and then the LSTM network is used to learn and correct the complex nonlinear structure in the modeling residuals. The experimental verification based on real logistics line data shows that the combined prediction model is superior to the traditional single model in trend fitting and error control, and can provide more accurate data support for resource allocation and scheduling decision-making in short-distance transportation.

Keywords

ARIMA; LSTM; cargo volume prediction; short-distance transportation; time series; combined model

References

1. [1] Zhou Haoyang, Ma Luyu, Dai Zeyu Emergency allocation and structural optimization of e-commerce logistics network packages [J]. Modern Business and Industry, 2024, 45 (20): 42-44
2. [2] Li Siqi, Yu Kun, Chen Yuhao Research on Resource Usage Prediction of High Performance Computing Platforms Based on ARIMA and LSTM [J/OL]. Computer Science, 2025:1-11 [2022-04-26] http://kns.cnki.net/kcms/detail/50.1075.TP.20250226.1825.010.html.
3. [3] Li Juan, Yu Jing, Li Dapeng Research on Unit Power Characteristics Based on Different Data Granularity [C]//The 10th China Wind Power Aftermarket Exchange and Cooperation Conference Dalian: [Publisher unknown], 2023: 267-274
4. [4] Hu Yankun, Wang Ning, Liu Shu, etc Research on the Application of Time Series Models and LSTM Models in Water Quality Prediction [J]. Small Microcomputer Systems, 2021, 42 (8): 1569-1573
5. [5] Weng Zixia Stock price analysis and prediction based on ARIMA model - taking China Construction Bank as an example [J]. Modern Information Technology, 2023, 7 (14): 137-141
6. [6] Chai Heping, Shen Xue, Ji Yingjie, etc Research on E-commerce Logistics Network Problems Based on Cargo Volume Prediction and Optimization Scheduling [J]. Science and Technology Innovation and Productivity, 2024, 45 (10): 22-27
7. [7] Wang Jian, Wang Jian, Xue Nianming, etc Security fence image segmentation technology based on hyperparameter optimization and lightweight deep neural network [J]. Modern Information Technology, 2024, 8 (16): 123-126
8. [8] Wang Anqi, Yi Min, Zhao Hanxue, etc River flood reverse flow calculation based on LSTM neural network model [J/OL]. Hydrology, 2025:1-11 [2022-04-26] https://doi.org/10.19797/j.cnki.1000-0852.20240148.
9. [9] Wang Lei, galloping horses, Qi Junyan, etc Prediction method for surface subsidence in mining areas based on ARIMA-LSTM [J]. Computer Engineering, 2025, 51 (1): 98-105
10. [10] Lin Yun Text classification model based on Transformer multi head attention mechanism [J]. Modern Information Technology, 2025, 9 (6): 100-104

About the authors

Liu Mengcong (2004-), female, Han ethnicity, from Baoding, Hebei Province, is currently pursuing a bachelor's degree. Her research interests include neural networks, deep learning, and pulse signal processing; Corresponding author: Guo Xiechao (1979-), male, Han ethnicity, from Fuping, Shaanxi Province, associate professor, PhD, research direction: deep learning, graph neural networks, and bioinformatics; Wang Weiren (2004-), male, Manchu, from Dandong, Liaoning Province, undergraduate studies, research direction: deep learning and sports training science.