Research On GPS Data Pre-Processing Method Based On Wavelet Analysis On RINEX Level And Its Program Realization

GPS data preprocessing is a very important step in precise positioning, whether or not preprocessing is done well, will affect final positioning accuracy and reliability directly. GPS is used by a wide variety of applications with varying needs for accuracy. It is still one of research hot points, that how to get accuracy positioning and reliability, under the environment affected by obstruction, electromagnetic and multi-path et al. Recently, many scholars have researched GPS data preprocessing based on wavelet analysis on how to detect and repair carrier beat phase cycle slip, de-noise carrier phase observations, extract multi-path signal from double difference phase readuals. Today, GPS data preprocessing based on wavelet analysis is used mainly in baseline solution. However, precise point positioning is developing tremendously, so it is necessary to research GPS data preprocessing methods based on wavelet analysis.In the thesis, the author use wavelet analysis as tool, researched how to improve GPS data quality of single station on RINEX level. There are three main parts in this paper, the first one is introduction to wavelet analysis foundation and GPS data preprocessing methods on RINEX level; the second one is introduction to GPSPPS software programming based on wavelet analysis; the last one is applications experiments used to improve PPP and baseline solution accuracy and reliability, using GPS RJNEX data preprocessed by GPSPPS. Specifically, main contents and conclusions of this paper are:1. We reviewed works on GPS data preprocessing based on wavelet analysis done by scholars at home and abroad, and discovered that most of recent researches focused on baseline solution using double difference phase observations. With the rapid development of PPP, it is more and more meaningful to research single station's GPS data preprocessing based on wavelet analysis. The author raised three preprocessing modes on RINEX level according to different kinds of signals. These three modes are preprocessing mode based on raw observations, based on linear combination observations and based on observation residuals of point positioning.2. We talked about preprocessing mode based on linear combination observations and mode based on observation residuals of positioning, and gave the formulas of how to adjust raw observations using residuals.3. Programmed preprocessing software GPSPPS, using object oriented method, dynamitic link library technology, developed in VC++ and MATLAB environment. Main frame of GPSPPS, file system, designation of main models and notes are introduced respectively.4. We did an experiment using observation data spanning 24 hours of DNCH station, on which will build one of permanent station of Crust Movement Observation Network of China in a few years. In the experiment, preprocessing mode based on raw observation was used; we got many new RINEX observation files using different wavelet analysis strategies, and then checked quality of new files using teqc software, then we used mpl and mp2 of teqc results to evaluate which wavelet analysis strategy is better. The results showed that noise structure influents mpl and mp2 obverse largely, sln is better than others; hard threshold is better then soft threshold; heursure threshold strategy is better compared to others.5. Another experiment was carried out using BJFS IGS station's data (day of year 298 and 349). In the experiment, we used all three preprocessing modes; preprocessed BJFS's data using different wavelet analysis strategies, and compared PPP accuracy between before and after preprocessing. The results showed that preprocessing modes based on raw observations and based on linear combination observations are not very ideal, while the mode based on observation residuals can get better accuracy.6. We introduced a new method to improve baseline solution accuracy and reliability by preprocessing both stations' observations of a baseline. We evaluated this method by an experiment. In the experiment, we first preprocessed both stations' observations of the baseline, and then processed the baseline using TGO software. Compared to the quality of baseline between raw data and preprocessed data, it showed that this method can improve baseline quality by choosing suitable preprocessing method and suitable wavelet analysis strategy. In the experiment, preprocessing carrier phase observations will reduce the baseline quality obviously, while preprocessing pseudo range observations will not. By choosing suitable wavelet analysis strategy may improve ratio of baseline solution, reduce reference variance, enable solute baseline successful while the original solution is failed.