Research On The In-motion Alignment Of Marine Strapdown Inertial Navigation System

Initial alignment of Strapdown inertial navigation system(SINS)is to provide initial attitude information for SINS,which is one of the key technologies affecting the SINS performance.The research on in-motion SINS initial alignment can make contributions to navigation field in terms of two aspects.For one thing,in-motion alignment can improve the alignment accuracy during vehicle motion.For another,in-motion alignment can break the limitation on the state of vehicle motion during alignment.Therefore,the research of in-motion alignment has high engineering value.Compared to the static-base alignment,in-motion alignment suffers from various error disturbances so that its principle and implementation is more complicated.Currently,in-motion SINS alignment still faces some challenges to be addressed.Among the in-motion alignment,coarse alignment based on gravity in inertial frame and fine alignment based on Kalman filtering(KF)have been widely studied and applied due to their advanced performance.In this thesis,these two alignment approaches have been deeply studied to further improve the comprehensive performance of the in-motion SINS alignment.Calculating the derivative of the gravity in inertial frame is the key to the coarse alignment based on gravity in inertial frame.To solve the ill-posed problem on the numerical differentiation,a coarse alignment approach is proposed based on the curve-fitting to the gravity data.According to the gravity data,a linear constraint least-square fitting of spatial circle is designed.Based on the fitting results,tangent vector of any data on the circle is derived to calculate the gravity derivative.By using the gravity and its derivative,the initial attitude can be determined.The proposed coarse alignment is a concise self-alignment approach,which can improve the accuracy an reliability of the SINS alignment.In the proposed coarse alignment,the errors of gravity data have bad effects on the accuracy of the circle fitting,which will decrease the alignment accuracy.Therefore,considering sensors noise and rocking motion,this thesis provides error analysis and study preprocessing methods to the error of gravity data.Wavelet threshold de-noising method is designed to effectively eliminated the random noise.Moving average filter is designed to smooth the errors caused by the accelerator bias.Besides,the errors from lever-arm are compensated directly by the mechanical formula of lever-arm acceleration.By using the preprocessing methods,the errors of gravity data can be removed to a large extent,so the alignment accuracy is improved.System observability analysis results can reflect the alignment accuracy and time in different conditions,so it is necessary before applying and designing KF-based fine alignment.A singular value decomposition method based on piece-wise constant system is applied to analyze the system observability in different conditions of vehicle motion and observation.Corresponding simulations have also been carried out to verify the observability analysis results.The observability analysis results can be used as theoretical basis for determining optimal fine alignment scheme.An optimal performance of Kalman filtering requires that the system model and the noise characteristic are precisely available.However,in practice,the uncertainty of the system model and noise always exist due to sensors noises,environment disturbances and operational error,etc.The uncertainty can decrease the filter accuracy even lead filter divergence.To simultaneously solve the problem of the model uncertainty and noise uncertainty,an adaptive robust Kalman filtering is proposed.Robust KF and Sage-Husa adaptive KF are studied respectively.Then,the combination of the two filter is effectively realized to generate the adaptive robust KF.The proposed filter can bound the variance of filtering error and estimate measurement noise in real time in the case of uncertainty.By applying the proposed filter to the SINS fine alignment,the alignment accuracy,stability and robustness can be improved.In this thesis,the improved approaches to the in-motion SINS coarse and fine alignment are proposed.Designing scheme and implementation are detailed,and simulation and experiments have been carried out to verify the performance of the proposed alignment.The experimental results have demonstrated that the preprocessing methods of gravity data can effectively remove data error to improve the accuracy of the circle fitting.The coarse alignment based on the circle fitting can be applied in various sea condition and provide high accuracy(the error of heading misalignment is less than 1 degree)within 5 minutes.Compared to conventional Kalman filter,the adaptive robust Kalman filter has better accuracy,stability and capability of resisting the noise disturbances.so the resulting fine alignment has satisfactory performance and high engineering value.