The application of optimal nonlinear/non-Gaussian filtering to the problem of inertial navigation system (INS) alignment is described. This approach is made possible by a new technique called particle filtering (PF). PF theory is introduced and nonlinear error equations of INS alignment on a stationary base in the case of large initial error angles are used. The algorithm for solving the problem of optimal estimation of the state vector described by nonlinear equations from linear measurements has been developed. The simulation results exhibit the superior performance of this approach when compared with classical suboptimal techniques such as extended Kalman filtering (EKF).
展开▼
