高精度光电稳定跟踪平台基准轴抖动或者缓慢漂移通常使得光纤陀螺(FOG)的输出信号中含有随机噪声。针对这一特点,通过对工程中实际采用的光纤陀螺实测数据进行时间序列分析,运用递推最小二乘法建立了噪声模型,并对其进行自适应卡尔曼(Kalman)滤波处理。通过Allan方差法分析结果表明,使用只对观测噪声协方差R阵进行自适应的Kalman算法滤波效果明显优于普通Kalman算法,且加入的计算量小,实时性能优于Saga-Huga自适应Kalman算法,对提高光电稳定跟踪平台性能有一定的实用价值。 High-precision Photoelectric Stabilization Tracking Platform Reference axis jitter or slow drift often results in random noise in the output signal of a fiber optic gyroscope (FOG). In view of this characteristic, the time series analysis of the actual measured data of the fiber optic gyroscope used in the project is carried out. The noise model is established by the recursive least square method and the adaptive Kalman filtering is performed. The results of Allan variance analysis show that the Kalman filtering algorithm which only adaptively covariance R covariance of observation noise is better than ordinary Kalman filter and the computational load is small and the real-time performance is better than that of Saga-Huga adaptive Kalman algorithm , Which has certain practical value for improving the performance of photoelectric stable tracking platform.