《计算机应用》唯一官方网站 ›› 2024, Vol. 44 ›› Issue (11): 3435-3441.DOI: 10.11772/j.issn.1001-9081.2023111705

• 数据科学与技术 • 上一篇    下一篇

基于多尺度特征融合的时间序列长期预测模型

刘文博1, 于连飞1, 谢冬梅1, 蔡闯1, 曲志坚1(), 任崇广1,2   

  1. 1.山东理工大学 计算机科学与技术学院,山东 淄博 255049
    2.济南浪潮数据技术有限公司,济南 250000
  • 收稿日期:2023-12-11 修回日期:2024-05-23 接受日期:2024-05-24 发布日期:2024-07-25 出版日期:2024-11-10
  • 通讯作者: 曲志坚
  • 作者简介:刘文博(1998—),男,山东菏泽人,硕士,CCF会员,主要研究方向:云计算、大数据分析
    于连飞(1998—),男,河南周口人,硕士,主要研究方向:云计算、大数据分析
    谢冬梅(1998—),女,山东淄博人,硕士,主要研究方向:云计算、大数据分析
    蔡闯(1999—),男,山东菏泽人,硕士,主要研究方向:云计算、大数据分析
    任崇广(1982—),男,山东临沂人,教授,博士,主要研究方向:云计算、人工智能。
  • 基金资助:
    山东省高等学校青年创新团队发展计划项目(2019KJN48)

Long-term prediction model of time series based on multi-scale feature fusion

Wenbo LIU1, Lianfei YU1, Dongmei XIE1, Chuang CAI1, Zhijian QU1(), Chongguang REN1,2   

  1. 1.School of Computer Science and Technology,Shandong University of Technology,Zibo Shandong 255049,China
    2.Jinan Inspur Data Technology Company Limited,Jinan Shandong 250000,China
  • Received:2023-12-11 Revised:2024-05-23 Accepted:2024-05-24 Online:2024-07-25 Published:2024-11-10
  • Contact: Zhijian QU
  • About author:LIU Wenbo, born in 1998, M. S. His research interests include cloud computing, big data analysis.
    YU Lianfei, born in 1998, M. S. His research interests include cloud computing, big data analysis.
    XIE Dongmei, born in 1998, M. S. Her research interests include cloud computing, big data analysis.
    CAI Chuang, born in 1999, M. S. His research interests include cloud computing, big data analysis.
    REN Chongguang, born in 1982, Ph. D., professor. His research interests include cloud computing, artificial intelligence.
  • Supported by:
    Youth Innovation Team Development Program of Shandong Provincial Higher Educational Institutes(2019KJN048)

摘要:

长期时间序列预测在多个领域中具有广泛的应用需求。但是,时间序列的长期预测过程中表现出的非平稳性问题是影响预测准确性的关键因素。为了提高时间序列长期预测精度,以及预测模型的普适性,构建了基于序列分解的多尺度融合注意力神经网络预测网络(MSDFAN)模型。该模型采用时间序列分解提取输入数据中的季节成分和趋势成分,对不同数据成分进行不同的预测建模,能够对具有多尺度稳定特征的非平稳时间成分进行建模和预测。实验结果表明,与FEDformer相比,MSDFAN在5个基准数据集上的预测结果的均方误差(MSE)和平均绝对误差(MAE)分别平均下降了12.95%和8.49%,MSDFAN模型在多变量时间序列上取得了更好的预测精度。

关键词: 长期预测, 深度学习, 序列分解, 多特征融合, 非平稳性

Abstract:

Long-term time series prediction has a wide range of application requirements in many fields. However, the non-stationarity problem shown in the long-term prediction process of time series is a key factor affecting the prediction accuracy. To improve the long-term prediction accuracy of time series and the universality of prediction model, a Multi?Scale Decomposition Fusion Attention Network (MSDFAN) was constructed. The model uses time series decomposition to extract seasonal components and trend components in the input data, and models different predictions for different data components, and is able to model and predict non?stationary time components with multi?scale stability characteristics. Experimental results show that compared with FEDformer, the Mean Squared Error (MSE) and Mean Absolute Error (MAE) of MSDFAN on five benchmark datasets are reduced by 12.95% and 8.49%, averagely and respectively. MSDFAN achieves a better prediction accuracy on multivariate time series.

Key words: long-term prediction, deep learning, series decomposition, multi-feature fusion, non-stationarity

中图分类号: