The predictions of Length-Of-Day (LOD) are studied by means of Gaussian Process Regression (GPR). The EOP C04 time-series with daily values from the International Earth Rotation and Reference Systems Service (IERS) serve as the data basis. Firstly, well known effects that can be described by functional models, for example effects of the solid Earth and ocean tides or seasonal atmospheric variations, are removed a priori from the C04 time-series. Only the differences between the modelled and actual LOD, i.e. the irregular and quasi-periodic variations, are employed for training and prediction. Different input patterns are discussed and compared so as to optimise the GPR model. The optimal patterns have been found in terms of the prediction accuracy and efficiency, which conduct the multi-step ahead predictions utilising the formerly predicted values as inputs. Finally, the results of the predictions are analysed and compared with those obtained by other prediction methods. It is shown that the accuracy of the predictions are comparable with that of other prediction methods. The developed method is easy to use.
Prediction of Length-of-day Using Gaussian Process Regression
The journal of navigation ; 68 , 3
2015
Aufsatz (Zeitschrift)
Englisch
Sunspot cycle prediction using Warped Gaussian process regression
Elsevier | 2019
|Prediction of Navigation Satellite Clock Bias by Gaussian Process Regression
British Library Conference Proceedings | 2015
|VEHICLE TRAJECTORY PREDICTION WITH GAUSSIAN PROCESS REGRESSION IN CONNECTED VEHICLE ENVIRONMENT
British Library Conference Proceedings | 2018
|A confidence-based roadmap using Gaussian process regression
British Library Online Contents | 2017
|