At present the precision of fiber optic gyroscope (FOG) is so lower that it is necessary to analyze
the performance of FOG and set up its drift error model to raise the precision. There are many noise
source in FOG, these noise sources and environment intrusion cause many random error terms, such as
bias instability, angular random walk,rate random walk and rate ramp. It is impossible to adopt general
analytical method (such as calculating mean and covariance) to confirm these random errors.
Now, the precision of FOG made in our country is low and to improve it cost high and is difficulty.
So this paper improve the system precision by software in the error modeling and filtering of the FOG
random drift. Now, the main method to minish the FOG random drift is Kalman filter. In this paper,the
FOG drift data is processed by Kalman filter,and the effect of filtering is analyzed. The simulation result
show that Kalman filter can minish the FOG random drift more simply and more efficiently.
Because the Kalman filter is based on the steady time series model of FOG random drift, in this
paper, FOG random drift data is validated to be a non-stationary time series, so the unsteady sample of
FOG drift needs statistical test and corresponding pretreatment using stochastic signal processing
methods, and then the mathematical model is establishing by time series analysis theory. It is proved that
the random noise can be represented by a single equivalent ARMA(auto-regressive moving average) or
ARIMA model that is simple to implement. The data pretreatment is made, the model is identified,the
method of using long autoregression method to estimate the coefficients are studied.
In the end, experimental modeling of the FOG random drift is carried out, the random noise of
FOG is processed by using Kalman filter. Experimental results demonstrate that the performance of the
filter is feasible and the model can reject the random noise of FOG.