EEMD、CEEMD算法与SVM在SST时间序列研究中的应用

发布时间：2018-04-22 07:00

  本文选题：海洋表面温度 + 经验模态分解　； 参考：《数学的实践与认识》2017年07期

【摘要】：海洋表面温度(SST)具有非线性、非平稳等特征,给处理和预测带来了很大的困难.将集合经验模态分解(EEMD)、改进的集合经验模态分解(CEEMD)与支持向量机(SVM)方法相结合,实现了对东北太平洋月平均海温距平序列(SSTA)的预测:首先应用EEMD或CEEMD方法将SST数据分解为多个本征模态函数(IMFs),然后应用SVM算法对各IMFs进行拟合、预测,最后对各IMFs预测结果叠加重构得到预测结果.EEMD-SVM和CEEMD-SVM数值模拟结果显示,预测最大误差小于0.25℃,并且CEEMD-SVM预测效果更好,为SST实际预测提供了参考.
[Abstract]:Ocean surface temperature (SST) has the characteristics of nonlinearity and nonstationarity, which makes it difficult to deal with and predict. This paper combines set empirical mode decomposition (EMD), improved set empirical mode decomposition (EMD) and support vector machine (SVM) method. The prediction of the monthly mean SST anomaly sequence in the Northeast Pacific Ocean is realized. Firstly, the SST data are decomposed into multiple intrinsic mode functions by EEMD or CEEMD, and then the IMFs is fitted and predicted by the SVM algorithm. Finally, the prediction results are obtained by superposition reconstruction of IMFs prediction results. EEMD-SVM and CEEMD-SVM numerical simulation results show that the maximum error of prediction is less than 0.25 鈩，

本文编号：1786087