A Method to Analyse and Eliminate Stochastic Noises of FOG Based on ARMA and Kalman Filtering Method
Xiaojing Li, Jiabin Chen, Yong Shangguan · 2014
This paper presents a method to model and analyze stochastic noises of fiber-optic gyroscope (FOG) in a strapdown inertial navigation system (SINS). Auto-regressive and moving average (ARMA) model was constructed using the method of time series. AR(3) model was best suited according to Akaike Information Criterion, while ARMA (2, 1) model was more often used in general engineering practise. In order to compare the difference of the two models, Kalman filtering algorithms were constructed specifically according to parameters of the two models. Amplitudes and Allan variances of the five FOG stochastic noises were calculated and compared to validate the effectiveness of Kalman filtering method. Simulation results show that Kalman filtering method can effectively eliminate stochastic noises of FOG, but that parameters of Kalman filtering which are related with the type of ARMA model should be regulated specifically.