Signal Extraction Method of GNSS Coordinate Time Series Based on EWT-MPE
Aiming at the problem of"annihilation"of some effective signals in residual noise in the process of GNSS coordinate time series signal extraction,a secondary signal extraction method(EWT-MPE)combining empir-ical wavelet transform(EWT)and multi-scale permutation entropy(MPE)is proposed.This method first decompo-ses the residual noise after the first filtering and generates empirical scale components and a series of empirical wavelet components,and then uses MPE as a threshold index to divide all component into information modal com-ponents,mixed modal components and noise modal components.The mixed modal components will then be extrac-ted again until there is no mixed component,and then the initial denoising signal and the information modal com-ponent are reconstructed into the final extracted signal,so as to achieve the purpose of the second time signal ex-traction.Through experimental research on the original data of 8 terrestrial network stations in China,the results show that the method is overall better than empirical mode decomposition(EMD)and EWT,with an average signal-to-noise ratio of 12.47 dB and an average energy percentage of 79.10%.Compared with EMD and EWT,the sig-nal-to-noise ratio is increased by 33.80%and 13.78%,and the energy percentage is increased by 16.15%and 5.87%respectively.It can effectively extract local trend features in the original signal and further improve the signal extraction effect.
万方数据
维普
