Abstract: In order to solve the problems about the noise of geocenter motion time seris, such as the complex types, strong randomness, and difficult to effectively separate from the signal and so on, the net translation method is used to process the weekly solution data from IGS station, which acquires the time series of geocenter motion from 2012 to 2018, and proposing a noise reduction method based on VMD and energy entropy. Firstly, the time series components from high frequency to low frequency in each direction are obtained by decomposing the time series of geocenter motion. Secondly, the energy entropy of each variational modal component is calculated to obtain the boundary between noise and signal. Then the signal component is reconstructed to obtain the time series of geocenter motion after noise reduction.Finally, comparing EMD and EEMD methods, it is shown that the method used in this paper has better effect on noise reduction from correlation coefficient, signal-to-noise ratio, residual energy percentage, variance contribution rate and other parameter evaluation indexes.

Key words: geocentric motion time series, noise reduction, variational mode decomposition, energy entropy, signal-to-noise ratio