The Error Processing And Trend Prediction Of GNSS Deformation Monitoring Coordinate Time Series In Strong Interference Environment

With the increasing application of GNSS in the field of deformation monitoring,the accuracy of GNSS monitoring time series has become an important influence factor of deformation stability analysis.In order to analyze the instability of deformable bodies more precisely,the appropriate technical methods are needed to preprocess the GNSS coordinate time series to obtain the more accurate information of deformation.These measures have important implications for the prevention of geological disasters such as landslides and debris flows caused by deformation.The deformation monitoring based on GNSS navigation system can not only record the value of the displacement of the real-time monitoring,but including the time,space and frequency information related to the deformation and displacement,which can be better for later data analysis services.However,the GNSS deformation monitoring is easy to be disturbed by the external environment,and its data sequence contains a large amount of coarse error,noise and other interference information,which is susceptible to cause misleading for the follow-up analysis.Therefore,in this paper,with an open-pit slope as the engineering background,combined with GNSS deformation monitoring coordinate time series,the data signal preprocessing and deformation trend prediction theory is studied in depth.The main research contents are as follows:(1)The 3 Sigma error elimination with wavelet analysis.On the base of the excellent time-frequency characteristics of wavelet analysis,the wavelet decomposition is used to extract the low frequency part of the deformation trend.At the same time,with 3 Sigma method of long data sequence recognition,processing,the improved 3 Sigma method of gross error detection and elimination is proposed based on wavelet analysis.The effectiveness of this method is proved by the simulation example and the actual case study.(2)Analysis and elimination of noise characteristics based on spectral index and improved semi soft threshold method.Taking into account of the influence factors of the GNSS coordinate time series noise,the noise characteristics of the data signals are analyzed by using the spectral index.According to the data signals with different noise characteristics,on the base of the existing soft and hard threshold method and semi soft threshold de-noising theory,the random characteristic data sequence is denoised by improved semi-soft threshold method and the traditional white noise sequence is denoised by the wavelet-based hybrid approach.(3)Time series prediction model of coupled grey system and fractal geometry.In order to predict the deformation trend of whole slope,this paper provides two different displacement prediction modeling method for three-dimensional space.A set of tests were put forward under the same experimental condition,results shows that the accuracy of monitoring point fitting model is much higher than space deformation prediction model.According to analysis in different part of the slope and engineering requirement,the universality of monitoring point fitting model for application was verified.On account of GNSS monitoring time series,through the analysis of the error sequence formation mechanism,combined with the rock and soil mechanics,statistics and geometric analysis and related theory,we carry out research from different aspects of GNSS deformation monitoring time series.So we can get scientific instructions for the reinforcement measures of the slope,and to achieve the effective control of landslide,collapse and other disasters.