GLOBAL IONOSPHERIC TOTAL ELECTRON CONTENT PREDICTION METHOD BASED ON DEEP RECURRENT NEURAL NETWORK

摘要

A global ionospheric total electron content prediction method based on a deep recurrent neural network. The method comprises a training stage and a prediction stage. The training stage comprises: 1, collecting global ionospheric total electron content heat maps, and forming an original picture sequence after a horizontal position is adjusted; 2, constructing a training sample set; and 3, constructing a global ionospheric total electron content prediction model based on a deep recurrent neural network, and training same by using the training sample set. The prediction stage comprises: 4, collecting K global ionospheric total electron content heat maps every day, and continuously collecting the heat maps for t days; adjusting the horizontal positions of pixels of each of the collected pictures, establishing prediction samples, and taking the prediction samples as the input of the global ionospheric total electron content prediction model to obtain predicted heat maps; and 5, carrying out longitude sorting on the predicted heat maps to obtain a predicted global ionospheric total electron content heat map. In the method, spatial and temporal changes in the ionosphere are combined, and existing observation data is fully and effectively used, thereby improving prediction precision.