Researches On The Theories And Algorithms Of Earth Orientation Paramters Prediction

The Variation of Earth Orientation Parameters is very complicated due to multiple excitations. The elaborate research of its regulation means a lot to the geodynamic. Meanwhile, the short term values of Earth Orientation Parameters, which cannot be obtained in real time, are needed for applications such as satellite navigation and deep space exploration. So, it is necessary to predict the Earth Orientation Parameters in short term. This dissertation mainly focuses on short term forcasting of Earth Orientation Parameters. The main work and contributions are summarized as follows:1. The main periodic term and its excitations are briefly introduced, so does the measurement of Earth Orientation Parameters. The robust estimation is used in the fitting and prediction of Earth Orientation Parameters. The results show that more reliable parameters can be obtained by robust least square estimation, and thus, the forecasting accuracy is improved.2. The application of ARMA model in the prediction of Earth Orientation Parameters is furtherly studied. It's discovered that , for the polar motion, it's more reasonable to describe it by ARIMA model. The polar motion parameters are forcasted by ARIMA model, and the result is compared with that of AR model.3. The feasibility of predicting Earth Orientation Parameters directly by nertral network in short terms is discussed. A combined model which takes the neutral network as function model and the ARMA model as stochastic model is proposed, and the corresponding results is compared with other's.4. The combination of Grey Model with time series or neutral networks is proposed to forecast the Earth Orientation Parameters. The result reveals that the grey model is convenient, and it can make fully use of the latest data. The prediction precision of combining model based on grey model is better than that of models based on least square estimation.