Temperature drift modeling and compensation of FOG combined extended forgetting factor recursive least square (EFRLS)

The performance of fiber optic gyroscope (FOG) exhibits temperature dependence when the temperature changes. In this paper, a two-stage compensation scheme is used for modeling the thermal-induced error for FOG with temperature ranging from −40 °C to 60 °C. First, a linear model based on Shupe coefficient is developed to compensate the deterministic error. Then, an extended forgetting factor recursive least square (EFRLS) combined with autoregressive (AR) process is proposed to compensate the residual drift. And the recursive least square (RLS) is used to tune the state transition matrix online. The Kalman filter (KF) and adaptive Kalman filter with double transitive factors (AKF-dtf) are also investigated to provide a comparison with this approach. Experiments with a single-axis FOG show the proposed method can be more effectiveness. E.g., the bias stability of FOG is 0.0770/h before compensation turns to 0.0080/h after compensation by this novel method compared to 0.0150/h by KF and 0.0110/h by AKF-dtf.