Research On The Prediction Theories And Methods Of Atomic Clock Bias For Navigation Satellites

Satellite clock bias(SCB)prediction plays an important role in satellite navigation system,such as optimizing clock parameters in navigation message,meeting the needs of real-time dynamic precise point positioning and providing the required information for satellite autonomous navigation.This paper mainly researches on theories and methods which can improve the accuracy and stability of prediction results for navigation satellite clock bias.The main achievements of this paper are listed as follows:1.It is conducted to study the prediction performance of several frequently-used methods used for navigation satellite clock bias prediction,including the quadratic polynomial model,spectral analysis model,the grey model,ARIMA model and Kalman filtering model.By using the SCB data of GPS,the prediction tests are carried out for each method to analyze their prediction results and then the advantages and disadvantages of these methods are compared and concluded.2.It is proposed that a wavelet neural network(WNN)model based on a once difference method to predict clock bias considering the characteristic of SCB.The main ideas are as follows: two SCB values of adjacent epoch firstly make once difference to obtain the corresponding once difference sequences,then modeling based on the sequences to predict once difference values of the following time series.At last,the predicted sequences are recovered to the corresponding predicted SCB.This method makes that the WNN model of SCB prediction is simple in structure and the predicting precision is higher.At the same time,a new data preprocessing method aiming at processing once difference sequence of SCB is presented.Finally,several predicting experiments are conducted to verify the new model and the results show that: the once difference method can evidently improve the prediction precision of SCB for the given WNN,and the new data preprocessing method can further improve the prediction performance of the WNN model for SCB predicting,and the results also turn out that the new model possesses better performance than the five common models in SCB prediction.3.Based on quadratic polynomial,a model of least-squares collocation to predict SCB is established by considering the random part of SCB.To determine the covariance function of this model,the paper selects the empirical function and combines the experiments of clock bias data comparing to the covariance function fitting of least squares collocation,then obtains reasonable value ranges of the parameter for the model's covariance function.At last,by concrete SCB prediction tests,the new method proves having good performance in the short-term prediction of clock bias and the value ranges of the parameter for the model's covariance function are reasonable.4.A new SCB prediction method based on the combination of gray system model and wavelet neural network is proposed in the paper.Firstly the gray model is used for fitting the given SCB values and the parameters of the gray model is also determined by these given values at the same time;and then fitting residuals of the gray model are used to model the wavelet neural network.Finally adding the prediction results of the wavelet neural network to the corresponding prediction results of the gray model,the predicted SCB values are obtained.By this means,the new method can improve the performance of the grey model in short-term prediction,and meanwhile it avoids constructing a complex network structure.In the end,using SCB data to conduct prediction tests,the results show that the proposed method can get better SCB prediction results.5.The paper proposes a distributed SCB prediction method based on the idea of combination prediction clock bias by using distributed computing technology which is paid more attentions in the field of information.Simultaneously,detailed technical routes and key steps are given.At last the paper analyses the feasibility and advantage of the new method.