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

With the launch of bds-3 system satellite,BDS system is developing steadily in a stable and fast direction,and will keep bds-3 / bds-2 satellites running together for a long time.As the time benchmark of satellite navigation,atomic clock is very important for positioning.Therefore,it is of great significance to study the characteristic analysis and processing of bds-3 new satellite,as well as the operation status of bds-2.At the same time,due to the instability of real-time clock offset products,satellite clock offset prediction is still a real-time or near real-time means that users can not give up.Based on this,this paper uses the BDS satellite clock offset data of ODTS,studies the preprocessing method considering the data quality of BDS system,evaluates the performance of some BDS satellite atomic clocks,proposed several clock offset prediction methods for BDS satellite clock offset data,and develops a software of BDS satellite.And the conclusion is as follows:(1)This paper introduces the main contents of BDS satellite clock performance index and characteristic analysis,describes the basic theory of satellite clock offset data preprocessing,and explains several main models of clock offset prediction.(2)Based on the analysis of BDS satellite clock offset data,the original data mapping shows that there are serious discontinuities and missing problems in BDS satellite data;phase data has jump phenomenon and frequency data is often abnormal;from the frequency data mapping,we can see that bds-3 satellite is more stable than bds-2;bds-2 phase modulation operation is more frequent than bds-3 system,and the amplitude of abnormal situation occurs Bigger.(3)In view of the problems in BDS clock offset data,especially the lack of original data and the poor quality of available data,a satellite clock offset data preprocessing and repair model based on quartile method and data quality weighting is proposed,and the advantages and disadvantages between the method and the conventional preprocessing method are compared.At the same time,the gross error of satellite clock offset is effectively detected and repair,and draw the results into a graph.(4)The characteristics of bds-2 and bds-3 satellites are analyzed from four main and intuitive perspectives.It includes frequency drift rate,frequency accuracy,stability and periodicity.The frequency accuracy of BDS satellite atomic clock is in order of magnitude,while the frequency drift value is mostly in order of magnitude,and the situation of no satellite is different.The frequency stability of BDS system satellite is in order of magnitude.From the perspective of satellite type,MEO satellite has better frequency drift characteristics,frequency accuracy and frequency stability than IGSO satellite.Because bds-3 satellites have good atomic clocks,its overall stability is better than bds-2.From the point of view of the periodicity of satellite clock offset,both bds-2 and bds-3 satellites have obvious periods.At the same time,the period length is related to the satellite operation period.(5)The application of grey neural network model in satellite clock offset is studied.The prediction characteristics and application range of series,parallel and embedded grey neural networks are analyzed.The series model is more useful when the modeling data is less,while the parallel model is better when the data is sufficient.At the same time,based on the advantages and disadvantages of several grey neural network models,an adaptive prediction model of satellite clock offset based on them is proposed.The experimental results show that the prediction accuracy of satellite clock offset is improved about 1% ~ 3%.(6)The clock error fitting residuals of bds-2 and bds-3 satellites are analyzed.A satellite clock error prediction model based on Tikhonov regularized noise reduction model is proposed according to the information of the residuals that have not been modeled,and the noise distribution of bds-2 and bds-3 satellites and the remediation effect of pollution noise are verified.Experiments show that the noise level of bds-2 satellite is more consistent with normal distribution,similar to random noise,while bds-3 satellite is more inconsistent,and there are more abnormal points.At the same time,the restoration effect was positively correlated with the normal distribution degree.However,the absolute value of noise of bds-3 satellite is small,which reflects the high quality of bds-3 satellite clock.(7)Based on the analysis of clock error fitting residual,a clock error prediction model based on lasso regularization noise reduction and pls residual processing method is proposed,and the improved median method is used to process the satellite clock error gross error.At the same time,Lasso regularization noise reduction method improves the compliance of satellite clock residual with normal distribution.Compared with the Tikhonov regularization method mentioned above,Lasso method removes some invalid modeling information of satellite clock offset,so the modeling efficiency is higher and the prediction effect is better.It is also verified by experiments that the processing of satellite clock error prediction residual based on PLS residual processing method is beneficial to the improvement of satellite clock error prediction accuracy.(8)Based on the preprocessing of clock offset data of BDS satellite,the characteristics of onboard atomic clock are analyzed,and the theories of clock offset prediction are proposed,the performance evaluation and prediction software of BDS satellite is developed,and its operation process and interface are displayed.The test results show that the software can process and analyze the satellite clock offset data effectively.