Research On Underwater Gravity Measurement Data Processing Method

Gravity aided navigation is one of the most important ways to achieve autonomous and hidden underwater navigation.With the development of technology,inertial sensors have reached very high precision.In this case,the further improvement of the precision of the gravity aided navigation system depends on the high precision gravity anomaly information.Therefore,the research of gravity aided navigation data processing method for underwater gravity measurement is of great significance.In this paper,the long-term Random Drift Modeling of gravity sensors,the filtering method of underwater dynamic gravimetry data,the processing method of external reference velocity measurement information,the correction algorithm of underwater gravity measurement and verification test based on measured gravity data of seawater have been researched and the details of the thesis are as following:(1)Long-term random drift modeling methods for gravimeters based on High-order time series modeling and Radial basis function(RBF)neural network modeling is studied.An AR(4)model was established for gravity sensor static measurement data.Based on the established static model,the Kalman filter method was used to compensate for the random drift error.A network learning algorithm that meets the modeling requirements of gravity measurement data is studied and long-term drift model experiments of gravity sensor static measurement data are constructed.What's more,an improved RBF neural network modeling method is proposed based on the parameters of the time series model.Experimental results show that this method does not require preprocessing of modeling data and is more efficient.(2)The nonlinear adaptive filter based on RBF neural network is studied.The simulation experiments are carried out for underwater gravity dynamic measurement data.The experimental results show that the Pre-filtering effect of the original measurement data is better than that of the Kalman filter.Secondly,the online modeling method based on Recursive least squares(RLS)algorithm is studied,and the H?filter is used for real-time filtering.The results of the filter experiment show that the H?filter has better robustness for real-time filtering of dynamic gravity anomaly data.(3)The principle of Doppler Velocity log(DVL)and the calculation method of DVL with four-beam Jenner configuration DVL are analyzed.Secondly,the errors of DVL velocity measurement are analyzed,including system error and velocity error.Finally,the DVL random measurement error is studied.The random error compensation method based on adaptive Kalman filter is proposed,and the speed maintenance algorithm based on?~2 test is proposed for the abrupt noise measurement.Simulation experiments show that the adaptive Kalman filter can effectively suppress the random noise of speed measurement,and the speed maintenance algorithm can effectively identify the abrupt noise.(4)The E?tv?s corrections is studied.Based on the influence of curvature radius error,the improved E?tv?s correction formula is given.According to the installation error between the gravity sensor and the gyro stabilization platform,the horizontal disturbance acceleration correction is studied..According to the difference of the vertical disturbance acceleration period,Butterworth filter are used to correct the high-frequency vertical acceleration,and the mean-value filter is used to suppress the low-frequency vertical disturbance acceleration.According to the navigation depth of underwater vehicles,the free space correction and interlayer correction are analyzed.(5)The underwater gravity measurement data processing method studied in this paper is applied to the processing of measured gravity data in a certain sea area.,and the accuracy of the repetitive lines and the accuracy of the intersections were evaluated.The experimental results show that the underwater gravity data processing method studied in this paper can effectively extract the gravity anomaly signal.