Research On Airborne Vector Gravity Data Processing And Error Compensation

Strapdown GNSS/INS combined airborne gravimetry system has the advantages of small volume,light weight and high reliability,and it can effectively acquire high-precision,medium and high resolution gravity field information.In this paper,the original observation data of the strapdown airborne gravimetry system developed by China is processed,and the gravity disturbance vector of the repeatlines is estimated,and its accuracy is evaluated.The main work and conclusions are as follows:1.Based on Newton's second law of motion,the mathematical model of strapdown airborne vector gravity measurement is deduced in detail,and the mechanical arrangement of inertial navigation system in local horizontal coordinate system and inertial coordinate system is given respectively.2.The mathematical models and error separation methods of airborne gravimetry data processing are introduced in detail,including Butterworth low-pass filtering,wavenumber correlation filtering,comparative linear correction and wavelet threshold filtering.3.In static state,the attitude angle of inertial navigation is divided into two stages.In the coarse alignment stage,the high frequency noise of inertial device output is firstly removed by the method of averaging low pass filtering,then the attitude angle of inertial navigation is roughly estimated by using the method of nine parameter least square.In the precise alignment stage,the attitude angle error of the inertial navigation system is estimated and corrected by using the kalman filtering method based on the speed updating.The calculation results show that the attitude angle converges faster in the initial alignment stage,and basically achieves stability in 3 s.After finishing the alignment,the estimated two stages of heading angle,pitch angle and roll angle are about 28 ",1.8" and 3.6 " respectively.4.In the local horizontal coordinate system,the kalman filtering method based on position and velocity update is used to process the GNSS/INS combined observation data.After the method of low-pass filtering is used for noise reduction,the gravity disturbance vector with certain resolution is obtained.Then,the horizontal and vertical components of the gravity disturbance of the repeatlines are compensated by applying Wavenumber Correlation Filtering(WCF)and linear correction of specific force and evaluated by using internal accuracy.The results show that after applying wavenumber correlation filter for the horizontal component of gravity disturbance of the six repeatlines,the average internal accuracy of the east and the north components is increased from 9.77 mGal,9.18 mGal to 5.95 mGal,3.83 mGal,respectively,with half-wavelength resolution of 7.5 km.After linear correction for the vertical component of specific force,WCF method is used to process the vertical component of gravity disturbance of the repeatlines.The average internal accuracy of the vertical component of gravity disturbance is improved from 1.27 mGal to 0.59 mGal.5.The kalman filtering method based on the acceleration update is used to estimate the vertical component of the gravity disturbance in the inertial coordinate system and compared with the vertical component of the gravity disturbance estimated in the local horizontal coordinate system.The results show that the accuracy of kalman filtering method based on the acceleration update in the inertial coordinate system is equivalent to the kalman filtering method based on the position and velocity update in the local horizontal coordinate system and the former algorithm is simpler.By using the two methods,the mean standard deviation of the gravity disturbance difference of the six repeatlines are 1.21 mGal and 1.27 mGal respectively with half-wavelength resolution of 7.5 km.For the same lines,the standard deviation of the gravity disturbance difference calculated by the two methods is 0.65 mGal.The gravity disturbance of the repeatlines obtained by the two methods have evident systematic error,which can be obviously reduced by linear compensation according to time.The maximum difference between the lines is reduced from 8.20 mGal,8.19 mGal to2.70 mGal,2.54 mGal respectively,and the mean internal accuracy is increased from 1.21 mGal,1.27 mGal to 1.06 mGal,1.10 mGal respectively.6.The improved wavelet threshold function is applied to filter the vertical component of the original gravity disturbance,and it is compared with the results of the method of forced denoising and the method of soft hard threshold denoising.The results show that the effect of denoising with improved threshold function is the best.When the wavelet is decomposed into ninth layers,the noise reduction effect of db8,db9 and sym7 wavelet is ideal,and the average inertial accuracy of disturbance gravity of three repeatlines can reach 1.05 mGal,1.15 mGal and 1.13 mGal,respectively,which is comparable to the result of Butterworth low-pass filtering with cut-off frequency of 0.004 Hz.