基于EMD和组合模型的太阳黑子时间序列预测

发布时间：2018-04-05 13:30

本文选题：太阳黑子　切入点：EMD分解　出处：《郑州大学学报(工学版)》2014年03期

【摘要】：太阳黑子是非线性、非平稳、多尺度变化的时间序列,且观测结果大多存在噪声的干扰.针对太阳黑子时间序列预测的复杂性,首先将原始数据序列通过小波去噪进行预处理,然后将去噪后的信号通过EMD分解产生若干个从高频到低频的IMF分量和余项.针对低频分量变化缓慢和高频分量波动性较大的特点,分别采用RBF神经网络模型和SVM模型进行预测,最后将各个分量的预测结果相叠加得到最终预测值.仿真结果表明,该模型具有较高的预测精度.
[Abstract]:Sunspots are nonlinear, non-stationary and multi-scale time series, and most of the observed results are disturbed by noise.In view of the complexity of sunspot time series prediction, the original data sequence is preprocessed by wavelet denoising, and then the de-noised signal is decomposed by EMD to produce several IMF components and remainder from high frequency to low frequency.In view of the slow change of low-frequency components and the large volatility of high-frequency components, the RBF neural network model and the SVM model are used to predict, respectively. Finally, the final prediction values are obtained by superposing the prediction results of each component.The simulation results show that the model has high prediction accuracy.
【作者单位】：郑州大学电气工程学院;
【基金】：河南省科技攻关计划项目(122102210102)
【分类号】：P182.41

【相似文献】