计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (12): 60-70.doi: 10.11896/jsjkx.241100011

• 数据库&大数据&数据科学 • 上一篇    下一篇

基于分解与集成的多尺度太阳黑子数量预测

赵宇轩1, 余定峰2,3, 李冬雪1, 徐以东1, 李北明1   

  1. 1 哈尔滨工程大学烟台研究院 山东 烟台 264000
    2 海洋电磁探测与控制湖北省重点实验室 武汉 430064
    3 武汉第二船舶设计研究所 武汉 430064
  • 收稿日期:2024-11-04 修回日期:2025-02-11 出版日期:2025-12-15 发布日期:2025-12-09
  • 通讯作者: 余定峰(DingfengYu@whu.edu.cn)
  • 作者简介:(zyx8986@hrbeu.edu.cn)
  • 基金资助:
    国家自然科学基金(52101383);中央高校基本科研业务费专项资金(3072024JJ2702,3072023LH0802)

Multiscale Sunspot Number Forecasting Based on Decomposition and Integration

ZHAO Yuxuan1, YU Dingfeng2,3, LI Dongxue1, XU Yidong1, LI Beiming1   

  1. 1 Yantai Research Institute of Harbin Engineering University, Yantai, Shandong 264000, China
    2 Hubei Key Laboratory of Marine Electromagnetic Detection and Control, Wuhan 430064, China
    3 Wuhan Second Ship Design and Research Institute, Wuhan 430064, China
  • Received:2024-11-04 Revised:2025-02-11 Published:2025-12-15 Online:2025-12-09
  • About author:ZHAO Yuxuan,born in 2002,postgra-duate,is a member of CCF(No.V1529G).His main research interest is time series analysis and prediction.
    YU Dingfeng,born in 1986,Ph.D,senior engineer.His main research interests include electromagnetic theory and its application and electromagnetic field signal processing.
  • Supported by:
    This work was supported by the National Natural Science Foundation of China(52101383) and Basic Research Funds of Central Universities (3072024JJ2702,3072023LH0802).

PDF (PC)

26

摘要： 太阳活动直接影响日球层环境和地球上的生命,太阳黑子数(SN)是最重要和最常预测的太阳活动指数之一。提高SN预测精度可以为气候模型提供更可靠的数据支持,对于理解太阳活动周期具有重要意义。对此,提出一种结合自适应噪声完备集合经验模态分解(CEEMDAN)、混合神经网络和注意力机制的多尺度SN序列预测模型。该方法使用3种不同的数据集,分别是1818－2024年每日SN、1749－2024年月均SN和1700－2023年年均SN。由于SN序列的非平稳性、非高斯性和非线性性质,因此先利用CEEMDAN将太阳活动在各时间尺度上的变化分量分解为若干不同频率子序列,将子序列与原始序列相结合作为强化特征集,增强模型对太阳活动变化的表征能力,再利用时序卷积神经网络(TCNs)作为特征提取的前沿,融入双向长短时记忆神经网络(BiLSTM)捕捉时间序列的长期依赖性,同时引入注意力机制(Attention)动态识别并加权序列中的关键时间特征。在3种数据集上进行消融实验,结果表明,所提模型各模块之间具有良好的协同作用。在此基础上对比部分已有模型,各数据集的预测精度均有所提高。利用该模型预测SN,得到年、月、日3种不同频率的预测结果,将预测结果作为多时间尺度特征融合形成最终预测结果。结果表明,太阳活动在2025年呈现出显著增强的趋势,并预计将在本年达到第25个太阳活动周期的活动高峰,年均SN峰值预计为233.9。

关键词: 太阳黑子数量预测, 时间序列分解, 时序卷积网络(TCN), 双向长短时记忆网络(BiLSTM), 注意力机制, 太阳活动周期

Abstract: Solar activity exerts a direct influence on the heliosphere environment and life on earth.SN represents one of the most crucial and frequently predicted indices of solar activity.Enhancing the accuracy of SN predictions can provide more reliable data support for climate models,which is of great significance for understanding the solar activity cycle.This paper proposes a multi-scale SN sequence prediction model,which combines Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN),a hybrid neural network and an attention mechanism.The method employs three distinct datasets:the Daily Total SN from 1818 to 2024,the Monthly Mean Total SN from 1749 to 2024,and the Yearly Mean SN from 1700 to 2023.Given the non-stationary,non-Gaussian and non-linear nature of the SN time series,the CEEMDAN method is initially employed to decompose the components of solar activity changes on various time scales into a number of sub-series with different frequencies.These sub-series are then combined with the original series as a reinforced feature set,thereby enhancing the model’s ability to characterize the changes in solar activity.TCNs are then employed as the primary means of feature extraction,followed by the incorporation of BiLSTM to capture the long-term dependence of the time series.Additionally,Attention mechanisms are introduced to dynamically identify and weight the key temporal features in the sequence.Ablation experiments are conducted on three datasets concurrently,and the results demonstrate a notable synergy between the modules of the proposed model.A comparison of the existing models on this basis reveals that the prediction accuracy of each dataset has been enhanced.The proposed model is employed to predict SN,resulting in the acquisition of three distinct frequencies of yearly,monthly,and daily prediction outcomes.These are subsequently integrated as multi-timescale features to generate the final prediction results.The results indicate that solar activity has shown a significant increasing trend in 2025 and is expected to reach the peak of Solar Cycle 25 within this year,with an estimated annual mean SN of 233.9.

Key words: Prediction of sunspot number, Time series decomposition, Temporal convolutional network, Bidirectional long short-term memory network, Attention mechanisms, Solar cycle

中图分类号: 

  • TP391
[1]TANG J,ZHANG X.Smoothing monthly mean prediction of Sunspot number based on Chaos theory [J].Acta Physical Sinica,2012,61(16):543-549.
[2]WANG J L,HAN Y B.Discussion on the method of “similar week” and the prediction of the month-by-month value of sunspot number in the 23rd week[J].Journal of Space Science,2000(3):278-281.
[3]MIAO J,WANG J L,LIU S Q,et al.Prediction of the start time of the 24th active week by using similarity cycle method [J].Chinese Journal of Space Science,2007(6):448-452.
[4]PAN Y,ZHANG X Y,ZHANG Q Z.Chaotic Time Series Prediction of Sunspots Number with Wavelet Packet-wavelet Neural Network[J].Computer Science,2013,40(6):260-264.
[5]DING G,ZHONG S S.Prediction of Sunspot number based on time-varying threshold process neural network [J].Acta Physical Sinica,2007(2):1224-1230.
[6]DING L G,LAN R S,JIANG Y,et al.Based on neural network of sunspots area smooth month average prediction [J].Journal of Atmospheric Sciences,2012,35(4):508-512.
[7]SUN T L,LI G H.EEMD and RBF neural network for sunspot monthly mean prediction[J].Computer Engineering and Applications,2017,53(24):252-256,262.
[8]LIANG B,LIN Y Q,DAI W,et al.Analysis of sunspot activity prediction based on multivariate LSTM network[J].Astronomical Research and Technology,2020,17(3):322-333.
[9]WANG Q,LI J,GUO L.Solar cycle prediction using a longshort-term memory deep learning model[J].Research in Astronomy and Astrophysics,2021,21(1):121-128.
[10]PALA Z,ATICI R.Forecasting sunspot time series using deep learning methods[J].Solar Physics,2019,294(5):50.
[11]PANIGRAHI S,PATTANAYAK R M,SETHY P K,et al.Forecasting of sunspot time series using a hybridization of ARIMA,ETS and SVM methods[J].Solar Physics,2021,296(1):6.
[12]KUMAR A,KUMAR V.Stacked 1D Convolutional LSTMsConvLSTM1D) Model for Effective Prediction of Sunspot Time Series[J].Solar Physics,2023,298(10):121.
[13]PENG X.Sunspot Activity and Cycle Prediction Study Based on Time Series Algorithm and CNN-LSTM Modeling[C]//2024 IEEE 4th International Conference on Power,Electronics and Computer Applications(ICPECA).IEEE,2024:1211-1215.
[14]ZHUO R,HE J,DUAN D,et al.Prediction of solar activities:Sunspot numbers and solar magnetic synoptic maps[J].Science China Earth Sciences,2024,67(8):2460-2477.
[15]BAI S,KOLTER J Z,KOLTUN V.An empirical evaluation of generic convolutional and recurrent networks for sequence mo-deling[J].arXiv:1803.01271,2018.
[16]SIAMI-NAMINI S,TAVAKOLI N,NAMIN S A.The Perfor-mance of LSTM and BiLSTM in Forecasting Time Series[C]//2019 IEEE International Conference on Big Data.2019:3285-3292.
[17]VASWANI A,SHAZEER N,PARMAR N,et al.Attention is all you need[C]//Proceedings of the 31st International Confe-rence on Neural Information Processing Systems.2017:6000-6010.
[18]ZHANG Y,YAN J.Crossformer:Transformer utilizing cross-dimension dependency for multivariate time series forecasting[C]//The Eleventh International Conference on Learning Representations.2023.
[19]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.2021:11106-11115.
[20]EKAMBARAM V,JATI A,NGUYEN N,et al.Tsmixer:Lightweight mlp-mixer model for multivariate time series forecasting[C]//Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining.2023:459-469.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路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