The fiber optic gyroscope (FOG) is the core component of the inertial system, which has been widely utilized in aviation, maritime, automotive, and other fields. The scale factor is one of the critical parameters to characterize the dynamic performance of the FOG. Due to the temperature sensitivity of the optoelectronic devices, the scale factor errors will be caused with the temperature change. The traditional compensation method usually compensates the output of the FOG in a stable working state. Without considering the difference between the characteristics of the FOG in the start-up stage and the stable working state, the compensation accuracy of the traditional method cannot meet the demand for quick start-up in application areas. According to the above analysis, a novel scale factor error compensation method based on the grey wolf optimization (GWO) and gated recurrent unit (GRU) neural network is proposed in this paper. Furthermore, a new experiment is designed to calibrate the scale factor error in the start-up stage, which can reduce the calibration time and ensure the calibration accuracy. The experimental results demonstrate the proposed method significantly improves the compensation accuracy compared with other methods, such as the multilayer perceptrons (MLPs) and the traditional GRU network.
Read full abstract