Analysis Of Atomic Clock Performance And Study On Clock Difference Prediction Algorithm

As the core component of satellite navigation system,the performance of atomic clock directly determines the positioning and timing service quality of satellite navigation system.In order to deeply understand the performance of GNSS time service,this paper starts with the deficiencies of the atomic clock performance evaluation theory and clock error prediction algorithm,preprocesses the precision clock error data of the atomic clock,compares and analyzes the performance of the spaceborne atomic clock in different navigation systems.At the same time,the grey BP neural network combined clock error forecasting model is improved and verified by examples.The main research contents and conclusions are as follows:In view of the in orbit performance of GNSS atomic clock,the precision clock difference data provided by Wuhan University data center are analyzed.GNSS atomic clock clock error data has the phenomenon of missing data throughout the day.The data of GPS and Galileo atomic clock clock difference are relatively stable,and data mutation exists in BDS clock C07 and Galileo satellite clock E19.The frequency stability of atomic clock is calculated by Hadamard variance,and the performance of atomic clock of different satellite navigation systems is compared.It is found that the performance of GPS and Galileo is better than that of BDS and GLONASS.Among them,GLONASS satellite clock frequency accuracy is the best,GPS and Galileo satellite clock frequency accuracy are better than BDS system;Galileo satellite clock frequency drift rate is the smallest,followed by GPS and GLONASS,China's BDS satellite clock frequency drift rate is the highest;Galileo atomic clock frequency stability is the best,followed by BDS and GLONASS,GPS satellite clock frequency stability is the worst.Setting the weight of background value as a variable for dynamic optimization and improving the initial value,the prediction accuracy of the improved Grey model is improved by 10%?20%;constructing the BP neural network model optimized by improved particle swarm optimization algorithm to improve the stability of the clock error prediction model;finally,the improved Grey model and the optimized BP neural network model are used to predict respectively,and the linear weighted combination prediction is obtained The final forecast results.The improved grey BP neural network prediction model improves the prediction accuracy by 20%?50%and the average stability by 20%.