Temperature drift compensation of fiber strapdown inertial navigation system based on GSA-SVR

Abstract

The fiber strapdown inertial navigation system (INS) has been widely utilized in aircraft, autonomous driving, robotics, and other fields due to the rapid growth of fiber technology. The core components of INS, the fiber optic gyroscope (FOG) and quartz flexible accelerometer (QFA), are prone to affected by the environmental temperature variation. The traditional temperature drift compensation method requires a considerable time to calibrate each inertial sensor individually. But after the six inertial sensors are integrated, the temperature environment inside the INS will become more complex. As a result, this device-level method has an unsatisfactory effect in the practical application of INS. Furthermore, the traditional method ignores the fact that the temperature drift characteristics of FOG are not the same at different working states. To solve the above problems, we propose a novel temperature drift compensation method of INS based on gravitational search algorithm (GSA) tuning support vector regression (SVR). The startup state and stable working state are modeled independently based on the FOG's features. Without adding additional thermometers to the inertial sensors, the temperature field is constructed by making full use of the information provided by the built-in thermometer of six inertial sensors. The experimental results verify the effectiveness of our method during different working states. And compared with the traditional polynomial fitting method, our method has better performance in the navigation experiment, the navigation accuracy increased by more than 50%.