Research On Nonlinear Filtering Algorithm For Underwater AUV Navigation

Accompanied by the development of underwater acoustic communication,energy,control and navigation technology,autonomous underwater vehicles are widely used in marine warning,underwater search,marine resource exploration,submarine topography survey and underwater environment monitoring and other fields.High-precision navigation system is the key technology for autonomous navigation in AUV,existing AUV navigation technologies include inertial navigation,acoustic navigation,and geophysical navigation.With the rapid development of acoustic location technology in recent years,acoustic navigation overcomes the disadvantage that inertial navigation will accumulate over time in inertial navigation,and is widely used in AUV navigation systems.Due to the relatively complex marine environment,the observation data has gross errors and the statistical characteristics of noise are unknown,which leads to a decrease in the filtering positioning accuracy.The study of adaptive filtering algorithms has become the key to solving the above problems.In addition,the standard kalman filter is a linear filter,and it is also very important to study the nonlinear filter algorithm for the nonlinear motion state problem of AUV.In response to the above issues,this article has conducted the following research:1.The motion state of underwater AUV affects the nonlinearity of the system differently.The accuracy and stability of the three filtering algorithms EKF?UKF and PF are analyzed for different nonlinearities of the dynamic system.The results show that the UKF algorithm is superior to the EKF and PF algorithms for strongly nonlinear Gaussian systems.2.In response to the sudden change in the AUV movement state,based on the basic principle of the kalman filter algorithm,the variable-dimensional kalman filter algorithm is used to achieve an accurate estimation of the motion state,and the accuracy of the underwater target dynamic positioning is improved.3.Aiming at the problem of declining positioning accuracy due to gross errors observed in AUV acoustic navigation systems,an improved UKF method of robustness is proposed.The algorithm changes the detection conditions of the gross errors,combines the two-stage function model and the measured noise scale factor to construct the robustness factor function,and reduces the filter gain with observed gross errors,thereby reducing the influence of the observed gross errors on the filtering results.The results show that the improved robust UKF algorithm can better control the influence of the gross errors in the observations on the filtering results.By constructing the robustness factor function,the accuracy and stability of the filtering results can be effectively improved.4.For the case that the statistical characteristics of system noise and measurement noise in AUV acoustic navigation system are unknown,the adaptive unscented Kalman filter algorithm based on Sage-Husa is studied.An adaptive robust UKF algorithm is proposed.The method simultaneously estimates the system noise and measurement noise,and ensures the positive definiteness of the noise matrix,while resisting the influence of the gross errors on the solution results.The results show that the adaptive-adaptive UKF algorithm is an effective method to deal with the situation that the conventional unscented Kalman filter increases the filtering error due to the fact that the preset noise characteristics are not the same as the actual situation and there are gross errors in the observed values.