海洋学研究 ›› 2019, Vol. 37 ›› Issue (1): 9-14.DOI: 10.3969/j.issn.1001-909X.2019.01.002

• 研究论文 • 上一篇    下一篇

基于EEMD和ARIMA的海温预测模型研究

张莹1, 谭艳春*2, 彭发定1, 廖杏杰1, 余昱昕2   

  1. 1.广东海洋大学 数学与计算机学院,广东 湛江 524088;
    2.广东海洋大学 电子与信息工程学院,广东 湛江 524088
  • 收稿日期:2018-08-17 修回日期:2018-12-20 出版日期:2019-03-15 发布日期:2022-11-21
  • 通讯作者: *谭艳春(1980-),女,讲师,主要从事信号处理方面的研究。E-mail:pipihll@126.com
  • 作者简介:张莹(1982-),女,吉林四平市人,博士(后),讲师,主要从事数据挖掘及地学应用方面的研究。E-mail:zhangying198206@126.com
  • 基金资助:
    广东省普通高校重点科研项目“创新强校工程”资助(2018KTSCX091);广东省大学生创新创业训练计划项目资助(CXXL2019082);“海之帆”起航计划科技发明制作类项目资助(qhjhkj201806)

Study on time series prediction model of sea surface temperature based on Ensemble Empirical Mode Decomposition and Autoregressive Integrated Moving Average

ZHANG Ying1, TAN Yan-chun*2, PENG Fa-ding1, LIAO Xing-jie1, YU Yu-xin2   

  1. 1. School of Mathematics and Computer, Guangdong Ocean University, Zhanjiang 524088, China;
    2. College of Electronical and Information Engineering, Guangdong Ocean University, Zhanjiang 524088, China
  • Received:2018-08-17 Revised:2018-12-20 Online:2019-03-15 Published:2022-11-21

摘要: 类型丰富、时空分辨率高的海洋探测数据,为信号分解和机器学习算法的应用提供了可能。本文针对如何建立有效的海温预测模型这一问题,使用高时空分辨率的海表温度(SST)融合产品,引入信号处理领域的集合经验模态分解(EEMD)和机器学习领域的自回归积分滑动平均模型(ARIMA)。首先利用最适于分解自然信号的EEMD方法,将海温数据分解成多个确定频率的序列;再利用ARIMA分别对各个频率的序列进行预测,最后将各个序列的预测结果进行组合。该方法在丰富数据的支撑下,比以往直接使用海温数据所建立的预测模型精度更高,为更好地进行海温预测提供了新方法。

关键词: 集合经验模态分解, 机器学习, 自回归积分滑动平均模型, 海表温度

Abstract: Various types of ocean detection data with high spatial and temporal resolution provides possibilities for the application of signal decomposition and machine learning algorithms. In order to establish an effective sea surface temperature (SST) prediction model, the high time resolution of the SST and the fusion of products were used in this study, introducing the ensemble empirical mode decomposition (EEMD) in the field of signal processing and autoregressive integrated moving average model (ARIMA) in the field of machine learning. Firstly, the SST data were decomposed into several frequency sequences by EEMD method, which was the most suitable for decomposing natural signals. After that the ARIMA was used to predict the sequence of frequencies, and then the predicted results of each sequence were combined. Compared with the previous method of directly using SST data to build prediction model, this method achieves higher accuracy because of its abundant data, and provides a new way for better SST prediction.

Key words: Ensemble Empirical Mode Decomposition, machine learning, Autoregressive Integrated Moving Average Model, sea surface temperatures

中图分类号: