Research On Marine Gravity Data Processing Method Based On Optimum Parameters Estimation

Gravity aided navigation method is aiming at the problem of inertial navigation system error accumulated over time.It is a strictly passive system which can not only extend the inertial navigation system reset time and ensures the concealment of the underwater vehicle at the same time.The high-precision Marine gravity information is an important premise of gravity matching aided navigation system for accurate positioning information.Therefore,the research for Marine gravity data processing of gravity matching aided navigation system has extremely important significance.Traditional low-pass filter has the problem of time delay and the gravity signal distortion.According to the above problem,this thesis has carried on the thorough research based on estimation filtering theory,the details of the thesis are as following:(1)In low signal-to-noise ratio of Marine gravity measurement data preprocessing for noise reduction,this thesis compared and analyzed the low pass filter,kalman filter and wiener filter.First of all,the gravity signals and noise frequency characteristics is discussed and low pass filter with linear phase is designed;Secondly,equation of system based on the second order gauss-markov anomaly model was established and the kalman filter filter parameters is determined;Finally,in order to further suppress strong noise signal of the gravity measurement data,this thesis improves the frequency-domain wiener filter based on least mean-square error rule by introducing regularization factor for effective inhibition of strong noise in the gravity measurement data;Processing results based on the measured gravity data denoising simulation show that show that the frequency of regular wiener filter is best in the above three kinds of denoising method.It not only played a good inhibitory effect in suppressing high frequency noise,but also avoid filtering useful signal in signal-to-noise overlapping frequency.(2)The positive and negative kalman and H? filter based on the theory of time series modeling and estimation filter are studied.AR model of the Marine gravity measurement data based on time series analysis was established and based on this model the system state equation and observation equation was established;Positive and negative kalman filter and H? filter are used to process the Marine gravity measurement data by the analysis.Processing results show that the Positive and negative kalman filter overcomes the front of gravity data convergence accuracy is not high of the traditional kalman filter;H? filter can shows the details change trend of the gravity curve and overcoming the the precision model required of traditional kalman filter.(3)The improved adaptive kalman and regularized particle filter based on the second order approximation model and estimation filter is studied.The transfer function is deduced based on the principle of gravity sensor and second-order approximate model of the system and the system equation is established.Based on this model,adaptive kalman filter of making correction to system parameters by the changes in environmental noise is studied and improved.Kalman filter is used to improved adaptive filter by estimate information to prevent environmental noise mutation results in the decrease of filtering precision and even divergence.In order to avoid the error caused by linearization of nonlinear system,particle filter is studied and regularization resampling algorithm is introduced to overcome the sample degradation.The test results based on the measured gravity data show,The improved adaptive kalman filter not only makes corresponding parameter adjustment in order to improve the filtering precision,but also has good robustness;the particle filter eliminates the the error of linearization of nonlinear system,and real reflects the features of gravity information.(4)C#and Matlab mixed programming method is used to design the sea gravimeter data processing software based on the theory of the estimation filter Combined with the study of filtering method.The estimation filter method was tested on the software platform and the test result shows that the software is not only stable and effective operated but also convenient parameter can be adjusted to meet different application background.