Nonlinear Problems Of The Inertial Navigation System

In this paper, the intention is to focus on finding an in-flight INS alignment method. This method enables the INS starting to work only after coarse initial alignment, and then achieve the precision alignment making use of the velocity information while the INS is working. The proposed inertial navigation course is traceable, so that the current position can be calculated again after the precision alignment to get more accurate position.The conventional method is based on linearized error models. Recent approaches use directly the nonlinear kinematical equations of the rigid body motion. These methods lead to a simpler system structure as well as to smaller estimation error variances.The particular contributions of this paper are "free-heading" navigation and "dual platform" INS. The "free-heading" navigation adopts absolute reference. The absolute velocity in horizontal plane is calculated directly and the free-azimuth system is adopted. The dependency of INS to azimuth is weakened, so that the linear bound is enlarged. In contrast to the typical techniques used for INS, a "dual-platform" INS scheme is proposed. The algorithm adds an absolute platform as well as the platform maintained in horizontal plane. There is no need "to apply torque" to the absolute platform.The method that is used to compress the inertial measurement data is studied. An effective compression scheme is proposed, as well as an effective compensating arithmetic for flight path rebuild. The amount of data to be stored and the calculation load to rebuild the flight path is reduced considerably. The accuracy of the flight path rebuild algorithm has been proved the same as conventional navigation algorithm by simulation.An ASIC for calculation of direction-cosine matrix that is changed with the signals, pulses or PWM, from RLG or flexible gyro is designed by means of VHDL. The maximum frequency of input pulses can be as high as 1MHz and the maximum angle rate can be as high as 400°/sec while the resolving power is 1.57". The elements of the matrix are expressed as 64 bits fixed-point numbers with sign. The calculation precision is better than 1.08×l0^-19. The hardware architecture adopt ring-shift chain with width of 64 bits. The kernel consists of 46 adders of 8 bits, which make up a pipeline of 3 levels.Based on a restructured data representation, the covariance of a new variable that consists of measurement vectors is expressed as a linear combination of unknown parameters. Noise covariances are then estimated by employing a recursive least-squares algorithm. The recursion EKF is adopted. The proposed method requires no a priori estimates of noise covariances, provides consistent estimates. Simulation analysis are carried out for the proposed scheme. The results of the analysis show that a satisfactory performance can be achieved for typical cases.The error equations of platform INS in geographic coordinates is deduced over again. A complete system matrix of error equations on a moving base is obtained. The errors produced from some typical error source on a stationary base are modified. Some long-term items are complemented.