Improvement and Application of the Prediction Algorithm of Atomic Clock Combination Clock Difference

Abstract

The clock error prediction of satellite navigation system requires high precision and strong prediction stability. A combined clock error prediction model is designed in this paper. The background value weight is set to the dynamic optimization variable, and the initial value is improved to optimize the grey model (GM(1,1)). At the same time, the network weight and thresholds of BP neural network are determined by particle swarm optimization algorithm to optimize the prediction performance. Then, the optimized preliminary prediction results of different GM(1,1) are taken as training values and input into the optimized BP network to obtain the final prediction results. The simulation experiment is carried out by using International GNSS Service organization precise clock difference data and compared with grey forecasting model before and after improvement, the quadratic polynomial model, the BP neural network model, the grey neural network model and the forecasting model in this paper, respectively. The results show the advantages of the combined prediction model and meet the requirements of real-time and high-precision clock error prediction of satellite navigation system.