Research On Theory And Method Of Marine Precise Acoustic Data Processing

Marine positioning and navigation technology is an important support for marine ship navigation,marine resources and environment survey,marine comprehensive management,marine resources survey and marine economic construction.It can provide accurate,continuous and reliable position,time and speed information in all time domain and all airspace.Facing the strategic requirements of national integrated navigation,positioning and timing(PNT),underwater acoustic positioning and navigation overcomes the characteristics of rapid attenuation of electromagnetic wave in water,and realizes the navigation and positioning of underwater targets through acoustic signal propagation,which has become an important technical means of underwater PNT scene.However,the complex marine observation environment brings great technical challenges to high-precision marine acoustic positioning and navigation.Based on the above requirements,this paper systematically studies the theories and methods of marine precision acoustic data processing combined with the research status of marine acoustic positiong and navigation.At the same time,aiming at the construction of global ocean sound speed field,high gross error pollution observation quality control,precision error correction and parameterization,spatio-temporal correlation error modeling and compensation in underwater acoustic positioning and navigation,and the theoretical method research,algorithm improvement and experimental verification analysis are carried out.The main contents of the paper are as follows:(1)Sound speed is a key parameter affecting marine acoustic positioning and navigation.Aiming at the high-precision construction of sound speed field(SSF)in complex ocean,this paper proposes a global sound speed field construction method based on back propagation(BP)neural network.This method constructs the sound speed field model through multi parameter sound speed model and neural network algorithm.At the same time,considering the problem of sample correlation selection of neural network,a sound speed field construction method based on BP neural network considering sample correlation is further proposed.This method selects the learning sample of BP neural network by analyzing the correlation between sound speed and distance.The experimental results show that compared with the traditional sound speed field construction method,the BP neural network method considering sample correlation can significantly improve the accuracy and stability of sound speed field construction,and the average root mean square(RMS)of global ocean sound speed field construction is better than 0.4 m/s.(2)The underwater acoustic positioning methods including underwater zerodifference positioning based on one-way sound ray,underwater zero-difference positioning based on two-way sound ray,underwater single difference positioning based on one-way sound ray,underwater single difference positioning based on twoway sound ray,one-way sound ray tracing and two-way sound ray tracing are studied,and the applicability of the above six methods in seafloor datum points positioning is analyzed,Combined with the accuracy evaluation indexes of internal coincidence and external coincidence,the accuracy of the above six methods is analyzed.The experimental results show that the accuracy of the one-way observation model is the same as that of the two-way observation model for the circular navigation track,and the two-way observation model is better than the one-way observation model for the cross navigation track.Compared with the other five algorithms,the sound ray tracing algorithm based on two-way sound ray has higher accuracy.(3)Aiming at the problem that the systematic error and gross error in marine acoustic observation seriously affect the accuracy of acoustic positioning,a robust zerodifference Kalman filter(KF)method based on random walk process and equivalent gain matrix is proposed in this paper.In this method,the time-varying systematic error is regarded as a random walk process,and the systematic error and position parameters are estimated through KF,The equivalent gain matrix is constructed to realize robust estimation.Experiments show that the robust zero-difference Kalman filter method can improve the positioning accuracy by controlling the influence of systematic error and gross error without amplifying the observation random noise.(4)The key to the refinement of underwater acoustic function model is to effectively to eliminate the systematic error caused by the temporal and spatial variation of sound speed and sonar signal delay.Therefore,this paper proposes a robust zerodifference Kalman filter method based on two-step systematic error estimation.This method constructs the observation model of sound speed error and time delay deviation.Combined with KF,the first step systematic error and position parameters are calculated,and the observation residual is obtained.The parameterized time-varying sound speed error parameters are fitted through the residual,and the second step systematic error and position parameters are estimated in combination with KF.The experimental results show that this method significantly improves the three-dimensional positioning accuracy of deep seafloor datum points.The RMS of its observation slant range residual is better than 13 cm and the position inconsistency of the underwater datum points obtained by different tracks is better than 0.5 m.(5)The problems of unknown noise statistical characteristics and observation gross error inevitably exsit in underwater dynamic acoustic navigation,which leads to the decline or even divergence of filtering accuracy.In view of the above problems and considering the strong nonlinearity of underwater dynamic navigation observation,an underwater acoustic navigation method based on adaptive robust unscented Kalman filter(UKF)is proposed.This method realizes the on-line compensation of system noise and the effective control of gross errors through Sage-Husa estimation and Huber weight function respectively.It is verified by AUV long baseline navigation simulation experiment and underwater towed ultra short baseline navigation experiment.The experimental results show that the adaptive robust UKF algorithm can further reduce the influence of gross error while adjusting the system noise,and significantly improve the accuracy and stability of AUV acoustic navigation.(6)Long baseline acoustic navigation technology based on seafloor reference network is an important means of marine acoustic navigation.The key problem is how to model and correct the systematic error of acoustic navigation system.Therefore,this paper proposes an underwater acoustic augmented navigation method based on systematic error modeling.Based on the systematic error estimated by the seafloor reference network,combined with the error modeling method,a piece-wise systematic error correction model is constructed to compensate the systematic error of the traditional acoustic navigation observation model.The experimental results show that this method can effectively correct the influence of systematic error and improve the accuracy of underwater acoustic navigation.The three-dimensional RMS of acoustic navigation in the reference network is about 1 m.