ERP Short-term Prediction Based on ARMA
YE Xiu-son · Journal of Navigation and Positioning · 2014
In this article,the identification method of auto-regressive moving average(ARMA)model was given based on the characters of the earth rotation parameters(ERP)history data.Algorithm of long auto-regression method of white noise has been used to estimate parameters of ARMA model.The ERP history data processing effectively demonstrated that the computation taking fit residuals as white noise to estimate ERP parameters has features of more effective and simple.Simultaneously,another feature of this method is solving linear equations in the whole process and avoids the nonlinear operation.In order to reduce strong correlation of ERP history data,the time series' trend term and periodic term of the ERP were deducted first,and then difference computation was adopted to process residual sequence.Finally,short-term of the ERP was predicted based on ARMA model.Feasibility and correctness of above algorithms were verified effectively,ERP short-term prediction results are close to the International Earth Rotation and Reference Systems Service(IERS)products.