The F10.7 Solar Radio Flux Prediction Based On LSTM Neural Network

The precision of the 10.7 cm solar radio flux (F10.7) prediction is of great significance for the orbital prediction of low earth orbit satellites. In this paper, the long short-term memory (LSTM) neural network is used as the prediction method of F10.7 data. The design of the network model and the selection of parameters are introduced as well. Meanwhile, this paper evaluates the prediction performance of the model by using relevant indexes. The result demonstrates that it is a suitable method to predict the F10.7 data and show its tendency. Besides, the prediction effect of the LSTM neural network results in a better effect than which the traditional backpropagation neural network demonstrates.