Abstract:To overcome the problem that the accuracy of polynomial model decays quickly during the period of rapid satellite frequency variation, we develop a new prediction model based on LSTM neural network. The experimental results show that when the satellite clock is running normally and smoothly, the results of these two models are almost identical. When the output frequency of satellite clock changes rapidly, the prediction accuracy of LSTM neural network model is significantly better than that of polynomial model, and it can still provide high precision forecast results of clock offset.
WANG Wei,HU Caibo,ZHAO He et al. Application of LSTM Neural Network in Satellite Clock Prediction during Satellite Clock Frequency Deterioration[J]. jgg, 2023, 43(4): 369-373.