An Effective Ionospheric TEC Predicting Approach Using EEMD-PE-Kmeans and Self-Attention LSTM.

Total electron content (TEC) is one of the most important parameters of the ionosphere, which reflects the main characteristics of the ionosphere. Its accurate prediction plays an important role in improving the accuracy of GNSS navigation and ensuring the remote communication of radio. In order to solve the problems that the current combination of EMD and LSTM has a large amount of computation and insufficient mining of high-level information in TEC time series. This paper proposes an effective ionospheric TEC predicting approach using EEMD-PE-Kmeans and self-attention LSTM (EPKSL). First, ensemble empirical mode decomposition (EEMD) is used to stabilize the TEC time series, which solve the problem of endpoint effect and mode aliasing caused by empirical mode decomposition (EMD). Next, an PE-kmeans algorithm is proposed to cluster and reconstruct the intrinsic mode function (IMF) components generated by EEMD with similar complexity, which not only reduces the calculation amount of the prediction model, but also improves the prediction accuracy. Finally, the reconstructed sub-signals are fed into the model we designed for prediction. The experiments on the ionospheric TEC data of Sanya station show that the root mean square error (RMSE) of the proposed model gets 1.23 TEC units (TECUs), the mean absolute error (MAE) obtains 0.93 TECUs, and the R² achieves 0.982. Experimental results demonstrate that our proposed model has higher performance in terms of RMSE, MAE and R² compared with several typical prediction methods, while the prediction time is reduced by 32.8% compared with EMD-LSTM. [ABSTRACT FROM AUTHOR]