Random drift error of fiber-optic gyroscope（ FOG） is one of the key factors that affect precision of inertial navigation system（ INS）

and precision of INS can be effectively improved by compensating according to math model for datas of FOG random drift. Auto regressive moving average（ ARMA） model for gyro random drift is established on the basis of large numbers of experiments

meanwhile

method of long auto regressive model is used to solve model parameters

and optimize parameters. Experimental results prove that the FOG random drift model optimized by wolf pack algorithm（ WPA） is more accurate. Compared with traditional method of time series analysis

modeling precision is increased greatly. The method has a better reference value for improving precision of FOG random drift modeling.