Study On Initial Alignment Methods Of Strapdown Inertial Navigation System

One of the key technologies of SINS is the initial alignment, whose accuracy affects the precision of SINS and whose speed effects the response rapidity of the weapon. Kalman filter is effective when it is applied to the initial alignment of SINS. The main purpose of the initial alignment of SINS is to initialize the attitude matrix. Through the state space model Kalman filter estimates the initial misalignments and revises the attitude matrix.The precision and the convergence speed of Kalman filter are closely related to the observability of the state variables and the certainty of the state spacee model. Generally the velocity errors are selected as the observation variables of initial alignment, which results in bad observability. While the state equation of initial alignment is usually linear, that is deduced from the assumption that the three misalignments are all small enough. When the azimuth misalignment is large, the linear state equation can not well and truly describe the errors , propagation. The dissertation mainly focuses on the two problems above, analyzing and comparing different alignment methods.When the three misalignments are all small, the initial alignment of SINS assisted by the carrier phase of the double antenna GPS is proposed. The observation equation is deduced, which takes the error of carrier phase single-difference as the observation variables. The observability of the two methods, namely self-alignment taking velocity error as observation variables and GPS assistant alignment taking both velocity errors and carrier phase single-difference errors as observation variables, is analyzed by using singularity value decomposition. And the two methods are compared through simulations. The method of the observability degree calculation is simplified. The corresponding singular value of every state is pointed out by the vectors of the right decomposition matrix and then the observability degree is calculated.When the azimuth misalignment is large, the nonlinear alignment method is studied. Firstly, a new nonlinear filter method, UKF (Unscented Kalman Filter), is studied and improved. The TPs (Transformation Points) is furthermore updated in the time update process of UKF recursive estimation, which introduces the system...