The datum axis of high precision photoelectric steady tracking platform jittering and drifting slowly often makes the output signal of fiber optic gyroscope (FOG) contain random noise.Based on the characteristics mentioned above,time series analysis of the data actually measured from FOG which is actually applied in engineering is conducted.Noise model is established by using the recursive least squares method,and it is processed with adaptive Kalman filter.With an exhaustive analysis by using the Allan variance method,it is shown that the filtering effect of Kalman algorithm that simply adapts the observation noise covariance matrix R is much better than that of normal Kalman algorithm,and the real-time performance is better than that of the Saga-Huga adaptive Kalman algorithm.For the Kalman algorithm,the amount of calculation added is small.This work has some practical value to improve the performance of photoelectric stable tracking platform.