Urban freight volume prediction method based on time series decomposition and multi-dimensional feature fusion

CN122390634APending Publication Date: 2026-07-14DONGFENG LOGISTICS (WUHAN CO LTD

Patent Information

Authority / Receiving Office: CN · China
Patent Type: Applications(China)
Current Assignee / Owner: DONGFENG LOGISTICS (WUHAN CO LTD
Filing Date: 2026-04-30
Publication Date: 2026-07-14

AI Technical Summary

Technical Problem

Existing cargo volume forecasting methods have shortcomings in forecast accuracy, multi-dimensional feature fusion capabilities, city-specific modeling, and adaptive model updates, resulting in low forecast accuracy and difficulty in meeting the refined management needs of logistics companies.

Method used

A city-level freight volume forecasting method based on time series decomposition and multidimensional feature fusion is adopted. The freight volume time series is decomposed into trend, seasonal and residual components through STL decomposition. Combined with multidimensional features such as time, city, economy, industry and event features, city clustering is carried out and an adaptive forecasting model is established. A credibility assessment mechanism and a dynamic correction mechanism are introduced.

Benefits of technology

It significantly improved the accuracy of city-level freight volume forecasting to 85%~92%, reduced forecasting errors, provided city-specific modeling and model adaptation capabilities, and enhanced the interpretability of forecast results and the scientific nature of decision-making.

✦ Generated by Eureka AI based on patent content.

Patent Text Reader

Abstract

The application relates to the technical field of logistics demand prediction, and discloses a city-level freight volume prediction method based on time series decomposition and multi-dimensional feature fusion, which comprises the following steps: establishing historical freight volume time series of each city; calculating the correlation coefficients of each feature and the freight volume, and screening out key features; performing STL decomposition to decompose the historical freight volume time series into three independent components, namely a trend component, a seasonal component and a residual component; clustering the cities in China to identify a plurality of city clusters; respectively establishing prediction models for the three components, and reconstructing the prediction results through a weighted combination mode of adaptive weights to obtain freight volume prediction values; and calculating the confidence scores of the freight volume prediction values to give confidence intervals and risk levels. The city-level freight volume prediction method based on time series decomposition and multi-dimensional feature fusion effectively solves the problems of low prediction accuracy, single feature dimension, difficulty in adapting to city differences and lack of self-updating capability in the prior art.

Need to check novelty before this filing date? Find Prior Art