Performance Analysis Of BDS Atomic Clock And Study On The Prediction Of Clock Offsets

The successful networking of the BeIDou-3 Global Navigation Satellite System(BDS-3)marks the successful completion of the "three-step" development strategy of BDS system.In navigation and positioning,the time system directly affects the accuracy of positioning.As the basis of the time system of satellites,it is important to evaluate the performance of atomic clocks on board.The satellite atomic clock of BDS-3 adopts the new generation of rubidium clock and hydrogen clock.However,there are few performance analyses of the satellite atomic clock of BDS-3 at present,and most of them are not comprehensive.Therefore,this paper conducts a long-term characteristic analysis of the performance of the satellite atomic clock of BDS-3.Aiming at the low accuracy and instability of real clock error data,a clock error prediction model optimized by beetle swarm optimization(BSO)algorithm on BP neural network was introduced,and the accuracy and stability of several common clock error prediction models were analyzed.Finally,in order to analyze the influence of the clock bias predicted by the model on the positioning accuracy,the clock bias predicted by the model is substituted into the precise single point positioning.The main research contents and achievements of this paper are as follows.(1)The performance of atomic clocks on BDS was analyzed by using the clock bias data of one year after the successful networking of BDS-3.The research shows that bDS2 has more frequent phase adjustment than BDS-3,and some satellites of BDS-2 have more phase adjustment than IGSO and MEO.Atomic clocks in MEO orbit in BDS systems are tuned more often than in IGSO and GEO,BDS-3 is tuned more often than BDS-2,and most hydrogen clocks in BDS-3 have a straight frequency trend,while the rubidium clock has a downward slope frequency trend.The clock drift of IGSO satellite in BDS system has obvious periodicity compared with other satellites in BDS system,and the clock drift value of BDS-3 is increased by about 50%compared with BDS-2.The fitting accuracy of the BDS-3 atomic clock is about 58%better than that of the BDS-2 atomic clock,and the fitting accuracy of the hydrogen clock is significantly better than that of the new rubidium clock.Most atomic clocks in BDS systems contain more than two distinct period terms.The overlapped Hadamard variance was used to analyze the ten-thousand-second stability of BDS,and it was concluded that the ten-thousand-second stability of BDS-3 was 67.8%higher than that of BDS-2.(2)The optimization results of particle swarm optimization algorithm and beetle swarm optimization(BSO)algorithm are analyzed.The results show that BSO has stronger global search ability and better convergence accuracy than particle swarm optimization algorithm.Fifteen satellites were used to analyze the clock error forecast.The BSO-BP model has a small change in the forecast accuracy of different satellites,and has a strong forecast stability,and the forecast accuracy gradually increases with the increase of fitting time.Compared with LP,QP and GM models,the prediction accuracy of BSO-BP model was improved by 85.5%,81%,73.1%and 78.5%,58.3%,68.3%,respectively,when the 6h forecast was fitted for 0.5h and 1h.Compared with LP,QP and GM models,the stability of BSO-BP model is improved by 50%,50%,56.5%and 57.6%,33.7%,64.2%,respectively,when the 6-hour and 12-hour data are fitted according to the standard deviation analysis of short-term forecast of 10 groups of clock bias data.According to the analysis of different atomic clocks,the prediction accuracy of hydrogen clock and new rubidium clock is 80.4%and 68.5%higher than that of rubidium clock,respectively.Meanwhile,the prediction stability of hydrogen clock and new rubidium clock is obviously improved compared with that of rubidium clock.Therefore,the overall performance of the atomic clock on the BDS-3 is better than that on the BDS-2.Finally,the data of 6 days were used for verification,and the results were consistent with those before.(3)Based on the analysis of positioning accuracy,the residual values of BSO-BP model are significantly better than those of the other three models in terms of value and concentration,and the positioning accuracy of the residual values of GM model is relatively scattered,which is similar to that of WHU precision clock difference data.In addition,the positioning accuracy of BSO-BP model is better than that of the other three models,and it is slightly better than that of WHU precision clock bias data in the E and N directions,and better than WHU precision clock difference data in the U direction.Compared with LP,QP and GM models,the positioning accuracy of BSO-BP model in the E direction is improved by 43.2%,44.2%and 28.1%,respectively.Compared with LP,QP and GM models,the positioning accuracy of BSO-BP model in N direction is improved by 28.4%,50.2%and 45%,respectively.Compared with LP,QP and GM models,the positioning accuracy of BSO-BP model in U direction is improved by 14.1%,0.8%and 36.2%,respectively.Figure[49]Table[18]Reference[99]...