利用IGS中心提供的不同纬度的电离层TEC值，建立基于改进的集总平均经验模态分解(MEEMD)算法和Elman回归神经网络(ERNN)模型相结合的电离层TEC预报模型。实验结果表明，在低、中、高不同纬度采用本文方法预报5 d电离层TEC的预测值的均方根误差最优可达到0.96 TECu，相对精度最优达到95.4%，精度较EMD-ERNN模型及单一ERNN模型有显著提高。"/>
Prediction Models of Ionospheric TEC by MEEMD and Elman Recurrent Neural Network" />
In this paper, we combine modified ensemble empirical model decomposition (MEEMD) algorithm with Elman recurrent neural network (ERNN) to predict TEC by values of different latitudes provided by IGS center. At different latitudes which are under low, medium and high, the experimental results show that the smallest mean square errors of 5 days’ ionosphere TEC is 0.96 TECu and the best relative precision is 95.4%. Our model is better than the EMD-ERNN model and the single ERNN neural network model"/>
<div style="line-height: 150%">Prediction Models of Ionospheric TEC by MEEMD and Elman Recurrent Neural Network</div>
Abstract In this paper, we combine modified ensemble empirical model decomposition (MEEMD) algorithm with Elman recurrent neural network (ERNN) to predict TEC by values of different latitudes provided by IGS center. At different latitudes which are under low, medium and high, the experimental results show that the smallest mean square errors of 5 days’ ionosphere TEC is 0.96 TECu and the best relative precision is 95.4%. Our model is better than the EMD-ERNN model and the single ERNN neural network model
