计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (2): 170-179.doi: 10.11896/jsjkx.250100137

• 数据库&大数据&数据科学 • 上一篇    下一篇

D-LINet:融合双线性层与双向归一化的时间序列预测框架

耿海军, 李东鑫   

  1. 山西大学自动化与软件学院 太原 030006
  • 收稿日期:2025-01-22 修回日期:2025-03-29 发布日期:2026-02-10
  • 通讯作者: 耿海军(genghaijun@sxu.edu.cn)
  • 基金资助:
    国家自然科学基金(62472267)

D-LINet:Time Series Forecasting Framework Integrating Dual-linear Layersand Dual Normalization

GENG Haijun, LI Dongxin   

  1. School of Automation and Software Engineering,Shanxi University,Taiyuan 030006,China
  • Received:2025-01-22 Revised:2025-03-29 Online:2026-02-10
  • About author:GENG Haijun,born in 1983,Ph.D,professor,is a member of CCF(No.84802M).His main research interests include network security and network architecture.
  • Supported by:
    National Natural Science Foundation of China(62472267).

PDF (PC)

79

摘要： 时间序列预测在能源管理、交通流量和气象分析等多个实际场景中具有重要应用价值。然而,时间序列数据中存在的分布漂移(Distribution Shift)与长程依赖(Long-term Dependency)仍限制了传统方法与现有深度学习模型在长期预测中的表现。为此,提出了一种名为 D-LINet(Dual-Normalization and Linear Integration Network)的创新模型。该模型结合了Dish-TS(Distribution Shift in Time Series Forecasting)框架的分布归一化能力与线性映射的高效性,并采用双向归一化与双线性层的设计,有效缓解输入与输出空间的分布偏移,增强了对周期性与趋势性特征的捕捉能力。在多个真实数据集上对D-LINet的预测性能进行了全面评估。结果显示,在短期与长期预测中,D-LINet 的均方误差和平均绝对误差均显著优于主流模型(如 Transformer,Informer,Autoformer和DLinear)。此外,实验还探讨了输入窗口长度及先验知识的引入对预测性能的影响,为后续模型优化提供了重要指导。该研究针对复杂分布漂移问题提出了新的解决思路,并有助于提升时间序列预测的精度与稳健性。

关键词: 时间序列预测, 分布漂移, 双向归一化, 线性映射, 周期性与趋势性建模

Abstract: Time series forecasting plays a crucial role in various real-world applications such as energy management,traffic flow forecasting,and meteorological analysis.However,the presence of distribution shift and long-term dependency in time series data continues to limit the performance of both traditional methods and existing deep learning models in long-range forecasting.To address these challenges,this paper proposes an innovative model named D-LINet.The proposed model integrates the distribution normalization capability of the Dish-TS framework with the efficiency of linear mappings.By employing dual-direction normalization and dual-linear-layer designs,it effectively mitigates distribution shifts in both input and output spaces,while significantly enhancing the capture of periodic and trend-related features.A comprehensive evaluation of D-LINet on multiple real-world datasets demonstrates that,for both short- and long-term forecasting,D-LINet consistently achieves lower MSE and MAE compared to mainstream models such as Transformer,Informer,Autoformer and DLinear.In addition,experiments investigate the influence of input window length and the incorporation of prior knowledge on forecasting performance,providing valuable insights for subsequent model optimization.Overall,this study offers a novel solution to address complex distribution shifts,contributing to improved accuracy and robustness in time series forecasting.

Key words: Time series forecasting(TSF), Distribution shift, Dual normalization, Linear mapping, Periodicity and trend modeling

中图分类号: 

  • TP393
[1]KHAN Z A,HUSSAIN T,ULLAH A,et al.Towards efficient electricity forecasting in residential and commercial buildings:A novel hybrid cnn with a lstm-ae based framework[J].Sensors,2020,20:1399.
[2]LI Y,GAO Y,YAO Z X,et al.Intelligent traffic flow prediction for data scarcity scenarios[J].Journal of Software,2024,35(4):7018-7030.
[3]ANGRYK R A,MARTENS P C,AYDIN B,et al.Multivariate time series dataset for space weather data analytics[J].ScientificData,2020,7:227.
[4]CHEN W,YANG Y Y,LIU J Y,et al.A combination model based on sequential general variational mode decomposition method for time series prediction[J].arXiv:2406.03157,2024.
[5]XU S J,CHAN H.Forecasting the demand of the aviation industry using hybrid time series SARIMA-SVR approach[J].Transportation Research Part E:Logistics and Transportation Review,2019,122:169-180.
[6]JIA Y,LIN Y F,YU J,et al.PGN:The RNN’s new successor is effective for long-range time series forecasting[J].arXiv:2409.17703,2024.
[7]HOU H W,YU F R.RWKV T S:Beyond traditional recurrent neural network for time series tasks[J].arXiv:2401.09093,2024.
[8]HU A,WANG D K,DAI Y,et al.TimeCNN:Refining cross-variable interaction on time point for time series forecasting[J].arXiv:2410.04853,2024.
[9]SHI J,MYANA R,STEBLIANKIN V,et al.Explainable parallel rcnn with novel feature representation for time series forecasting[C]//International Workshop on Advanced Analytics and Learning on Temporal Data.IEEE Computer Society,2023:56-75.
[10]VASWANI A,SHAZEER N,PARMAR N,et al.Attention isall you need[C]//Advances in Neural Information Processing Systems.IEEE Computer Society,2017:5998-6008.
[11]DEVLIN J,CHANG M W,LEE K,et al.BERT:Pre-training of deep bidirectional transformers for language understanding[J].arXiv:1810.04805,2018.
[12]DONG L,XU S,XU B.Speech-transformer:A no-recurrence sequence-to-sequence model for speech recognition[C]//2018 IEEE International Conference on Acoustics,Speech and Signal Processing(ICASSP).IEEE Computer Society,2018:5884-5888.
[13]LIU Z,LIN Y,CAO Y,et al.Swin transformer:Hierarchical vision transformer using shifted windows[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.IEEE Computer Society,2021:10012-10022.
[14]ZENG A,JU X,YANG L,et al.DeciWatch:A simple baseline for 10x efficient 2D and 3D pose estimation[C]//17th European Conference on Computer Vision(ECCV).IEEE Computer Society,2022:607-624.
[15]ZHOU H,ZHANG S,PENG J,et al.Informer:Beyond efficient transformer for long sequence time-series forecasting[C]//Proceedings of the AAAI Conference on Artificial Intelligence.IEEE Computer Society,2021:11106-11115.
[16]XU J,WANG J,LONG M,et al.Autoformer:Decompositiontransformers with auto-correlation for long-term series forecasting[C]//Advances in Neural Information Processing Systems.IEEE Computer Society,2021:22419-22430.
[17]ZENG A,CHEN M H,ZHANG L,et al.Are transformers effective for time series forecasting?[C]//Proceedings of the AAAI Conference on Artificial Intelligence.IEEE Computer Society,2023:11121-11128.
[18]WANG R,ZHANG Y C,WANG W D,et al.Algorithm of mixed traffic scheduling among data centers based on prediction[J].Journal of Computer Research and Development,2021,58(6):1307-1317.
[19]SALINAS D,FLUNKERT V,GASTHAUS J,et al.DeepAR:Probabilistic forecasting with autoregressive recurrent networks[J].International Journal of Forecasting,2020,36(3):1181-1191.
[20]RANGAPURAM S S,SEEGER M W,GASTHAUS J,et al.Deep state space models for time series forecasting[J].Advances in Neural Information Processing Systems,2018,31:7785-7794.
[21]WEN Y M,LIU S,MIAO Y Q,et al.Survey on semi-supervised classification of data streams with concept drifts[J].Journal of Software,2022,33(4):1287-1314.
[22]FAN W,WANG P,WANG D,et al.Dish-ts:a general paradigm for alleviating distribution shift in time series forecasting[C]//Proceedings of the AAAI Conference on Artificial Intelligence.IEEE Computer Society,2023:7522-7529.
[23]Local climatological data[DB/OL].https://www.ncei.noaa.gov/data/local-climatological-data/.
[24]ZHOU H,ZHANG S,PENG J,et al.Informer:Beyond efficient transformer for long sequence time-series forecasting[C]//Proceedings of the AAAI Conference on Artificial Intelligence.IEEE Computer Society,2021:11106-11115.
[25]Electricity load diagrams 2011－2014[DB/OL].https://archive.ics.uci.edu/dataset/321/electricityloaddiagrams20112014.
[26]LOSHCHILOV I,HUTTER F.Decoupled weight decay regularization[C]//7th International Conference on Learning Representations.IEEE Computer Society,2019:1-11.
[27]DU S D,LI T R,YANG Y,et al.A sequence-to-sequence spatial-temporal attention learning model for urban traffic flow prediction[J].Journal of Computer Research and Development,2020,57(8):1715-1728.
[28]PASZKE A,GROSS S,MASSA F,et al.Pytorch:An imperative style,high-performance deep learning library[J].Advances in Neural Information Processing Systems,2019,32:8024-8035.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发