GNSS coordinate time series noise reduction combined with VMD and Bi-LSTM
For the problem that noise in GNSS coordinate time series is difficult to be effectively removed,this paper constructs a joint variational mode decomposition and bidirectional long short-term memory model to remove noise in GNSS coordinate time series. Firstly,the GNSS coordinate time series is decomposed into k eigenmode function components,and the effective modal components are selected according to the sample entropy,and then processed by the bidirectional long and short term memory network respectively. Finally,the signals are synthesized. Taking the coordinate data of 12 GNSS stations such as BJFS with long time series and good data integrity as an example,the coordinate time series is denoised. Compared with the traditional decomposition method,it is found that in the E,N and U directions,the correction rate of velocity uncertainty is increased by 11. 03％,4. 60％ and 7. 39％,respectively,compared with the single variational mode decomposition,and by 31. 70％,27. 70％ and 24. 42％,respectively,compared with the empirical mode decomposition. The results show that this method can remove the noise in the signal better than the traditional decomposition method,and can improve the reliability of the signal.
GNSS coordinate time seriesvariational mode decompositionbidirectional long short-term memorysample entropysignal denoising
万方数据
维普
