부산대학교 도서관

The 9-axis micro-attitude sensors include gyroscopes, accelerometers, and magnetometers, as medium and low precision sensors in robots, drones, smart munitions, and other fields widely used. The traditional error estimation and compensation method usually need to rely on the rotary table for calibration, which is difficult to migrate and adapt to the different application environments, so this paper uses a high-precision sensor for calibration, establishes a 9-axis micro-attitude sensor error mathematical model, and proposes an integrated intelligent error compensation method based on neural network, and compares and analyzes the compensation accuracy and compensation time. The experimental results show that compared with the traditional LS method, the time of the integrated intelligent error compensation calculation based on the neural network is shortened by 95%, and the accuracy of the gyroscope, accelerometer, and magnetometer is increased by up to 32.77%, 81.44% and 88.51% after the compensation, which makes up for the lack of accuracy relying on the sensor for calibration to a certain extent.