Time Series Forecasting Method Based on Multi-Scale Feature Fusion and Autoformer

Accurate time series forecasting is crucial in fields such as business, finance, and meteorology. To achieve more precise predictions and effectively capture the potential cycles and stochastic characteristics at different scales in time series, this paper optimizes the network structure of the Auto...

Full description

Saved in:
Bibliographic Details
Main Authors: Xiangkai Ma, Huaxiong Zhang
Format: Article
Language:English
Published: MDPI AG 2025-03-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/15/7/3768
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849738475390107648
author Xiangkai Ma
Huaxiong Zhang
author_facet Xiangkai Ma
Huaxiong Zhang
author_sort Xiangkai Ma
collection DOAJ
description Accurate time series forecasting is crucial in fields such as business, finance, and meteorology. To achieve more precise predictions and effectively capture the potential cycles and stochastic characteristics at different scales in time series, this paper optimizes the network structure of the Autoformer model. Based on multi-scale convolutional operations, a multi-scale feature fusion network is proposed, combined with date–time encoding to build the MD–Autoformer time series forecasting model, which enhances the model’s ability to capture information at different scales. In forecasting tasks across four fields—apparel sales, meteorology, finance, and disease—the proposed method achieved the lowest RMSE and MAE. Additionally, ablation experiments demonstrated the effectiveness and reliability of the proposed method. Combined with the TPE Bayesian optimization algorithm, the prediction error was further reduced, providing a reference for future research on time series forecasting methods.
format Article
id doaj-art-0f93c0018a384645869e71e28e42a04c
institution DOAJ
issn 2076-3417
language English
publishDate 2025-03-01
publisher MDPI AG
record_format Article
series Applied Sciences
spelling doaj-art-0f93c0018a384645869e71e28e42a04c2025-08-20T03:06:32ZengMDPI AGApplied Sciences2076-34172025-03-01157376810.3390/app15073768Time Series Forecasting Method Based on Multi-Scale Feature Fusion and AutoformerXiangkai Ma0Huaxiong Zhang1School of Information Science and Engineering, Zhejiang Sci-Tech University, Hangzhou 310000, ChinaSchool of Computer Science and Technology (School of Artificial Intelligence), Zhejiang Sci-Tech University, Hangzhou 310000, ChinaAccurate time series forecasting is crucial in fields such as business, finance, and meteorology. To achieve more precise predictions and effectively capture the potential cycles and stochastic characteristics at different scales in time series, this paper optimizes the network structure of the Autoformer model. Based on multi-scale convolutional operations, a multi-scale feature fusion network is proposed, combined with date–time encoding to build the MD–Autoformer time series forecasting model, which enhances the model’s ability to capture information at different scales. In forecasting tasks across four fields—apparel sales, meteorology, finance, and disease—the proposed method achieved the lowest RMSE and MAE. Additionally, ablation experiments demonstrated the effectiveness and reliability of the proposed method. Combined with the TPE Bayesian optimization algorithm, the prediction error was further reduced, providing a reference for future research on time series forecasting methods.https://www.mdpi.com/2076-3417/15/7/3768time series forecastingmulti-scale feature fusion networkdate–time encodingAutoformerBayesian optimization algorithm
spellingShingle Xiangkai Ma
Huaxiong Zhang
Time Series Forecasting Method Based on Multi-Scale Feature Fusion and Autoformer
Applied Sciences
time series forecasting
multi-scale feature fusion network
date–time encoding
Autoformer
Bayesian optimization algorithm
title Time Series Forecasting Method Based on Multi-Scale Feature Fusion and Autoformer
title_full Time Series Forecasting Method Based on Multi-Scale Feature Fusion and Autoformer
title_fullStr Time Series Forecasting Method Based on Multi-Scale Feature Fusion and Autoformer
title_full_unstemmed Time Series Forecasting Method Based on Multi-Scale Feature Fusion and Autoformer
title_short Time Series Forecasting Method Based on Multi-Scale Feature Fusion and Autoformer
title_sort time series forecasting method based on multi scale feature fusion and autoformer
topic time series forecasting
multi-scale feature fusion network
date–time encoding
Autoformer
Bayesian optimization algorithm
url https://www.mdpi.com/2076-3417/15/7/3768
work_keys_str_mv AT xiangkaima timeseriesforecastingmethodbasedonmultiscalefeaturefusionandautoformer
AT huaxiongzhang timeseriesforecastingmethodbasedonmultiscalefeaturefusionandautoformer