Wavelet Analysis For GPS Monitoring Deformation Data And Its Application

In allusion to the different noising distribution of GPS monitoring deformation data, the theoretical analysis has been carried on the wavelet conversion characteristics, wavelet vanishing matrix, regularity, tight nature, symmetry and so on. The selection of optimal wavelet basis is studied in the processing of pre-treating deformation data. On the basis of experimental analysis and practical application with different wavelet functions to the Gauss noise, the systematic characteristic unwanted signal and the sudden change signal which is included in the deformation observation data sequence, the new signal de-noising method of the max-mode and non-linear threshold algorithm in wavelet transformation are proposed. The error estimating algorithms of the threshold and the auto-adapting threshold are proposed also. According to the wavelet's multi-resolution or multi-criteria characteristics and the powerful approaching ability of artificial neural network, the wavelet-Kalman filtering model, the wavelet-artificial neural network model and its algorithms are founded and realized. Their respective advantages are fully displayed and unified. The non-linear deformation prediction is carried out. Finally, the GPS deformation monitoring data processing system in the level of RINEX is developed based on the VC++ language and MATLAB platform. It can provide a new processing technique for GPS deformation monitor data and forecast the deformation in practice successfully.