On-line modeling and filter of high-precise FOG signal

Based on the general time-sequence model of gyroscope random error, an improved ARMA model was designed, by which the high precise fiber optic gyroscope (FOG) random error model was built on-line. According to this model and the Kalman filter arithmetic, the FOG random error was filtered in real time in the process of initial alignment and navigation of FOG inertial navigation system. Filtering results and the Allan variance analysis prove that the angle random walk, the bias instability, the rate random walk, the angular rate ramp and the quantification noise of FOG are twice less than those before the FOG random error is filtered. So this modeling and filtering methods can reduce the high-precise FOG error and improve the FOG precision effectively.