Prediction Of Earth Orientation Parameters And Its Effect Research On Orbit Computation Of Navigation Satellite

Precise transformation matrix between the celestial reference frame and the terrestrial reference frame is needed for positioning,navigation and tracking on earth and in space,and the conversion relation between the celestial and terrestrial coordinates is described by EOP.Due to the complexity of the space-geodetic data processing,EOP is acquired with a delay of several days,which has a great impact on the need for real-time applications,making the prediction of EOP be worthy of study.Prediction of EOP is affected by a variety of geophysical excitation source,so the variety of prediction is very complicated.The dissertation focuses on the theories and algorithms of EOP prediction and the factors affecting EOP prediction accuracy,the prediction results is applied to orbit computation,then the impact of EOP prediction errors on navigation satellite orbit computation of different orbit type is analyzed.The main research works in this dissertation are as follows.(1)The characteristics of EOP series are analyzed and the models of EOP are formulated.The acquired way of EOP observation data is illustrated,and the formula to obtain the atmospheric angular momentum and ocean angular momentum data is derived.The autocorrelation and partial autocorrelation properties of the polar motion,UT1-UTC and LOD series data in IERS bulletin B are analyzed.The Fourier transform amplitude spectrum characteristics and the properties based on discrete Meyer wavelet basis function of EOP are studied.then combining the periodic characteristics of the polar motion,UT1-UTC and LOD by spectral analysis and the EOP curve,together with the autocorrelation analysis results,the models of polar motion,UT1-UTC/LOD are formulated.On the basis of the EOP models,the stationary test sequence,normality test and zero mean test results of model residuals are proposed.(2)Time series analysis prediction methods for EOP are studied and proposed.Direct estimation and recursive estimation methods for AR model parameters are presented,on this basis,the AR model prediction method is derived.The reduced-order MAR model parameter estimation method is developed,and the reduced-order MAR model prediction formula is derived.Accuracy evaluation error rule for EOP prediction is Defined,including AE,MAE and RMS precision evaluation form.The error precision evaluation criteria is introduced into the prediction of polar motion based on AR model and the prediction of UT1-UTC based on reduced-order MAR model considering wind terms of the atmospheric angular momentum,and the results of polar motion and UT1-UTC prediction can clearly get.The difference between the polar motion value calculated by atmosphere,oceans and integrated atmospheric and oceanic excitation function and the observed value is comparative analyzed,and two variables values affected the polar motion calculation are illustrated.(3)Neural network prediction methods for EOP are studied and proposed.The artificial neural network structure,transfer function and learning algorithm for BP neural network are discussed.The spectrum features of resampling polar motion series are analyzed,the method of determining the interpolated basic series of polar motion by spectral analysis is presented,A single output neural network structure is designed to predict the resampling polar motion data.A single output and a multiple output neural network prediction models for UT1-UTC are formulated,then the short-term prediction characteristics for UT1-UTC by neural network are discussed.(4)The impact of EOP prediction errors on orbit computation of navigation satellite is analyzed.The error propagation relationship between EOP and HG is given.The formula of the impact of EOP prediction errors on the acceleration of the Earth's non-spherical gravitational force is deduced.The influence of EOP prediction error on acceleration magnitude of navigation satellite and the influence of different order gravitational field on acceleration calculation are comparative analyzed,and the influence prediction errors of polar motion and UT1-UTC on orbit extrapolation and satellite ephemeris accuracy are studied.This dissertation enriches and expands the traditional EOP prediction methods and the orbit computation theory,which have some theoretical significance.The proposed linear and nonlinear prediction methods of EOP greatly improve prediction accuracy,which have high application value in real-time satellite tracking and navigation.