Low cost inertial sensor attitude fixation algorithm and accuracy analysis

Abstract

Aiming at the problems of angle oscillation, slow convergence speed, and low accuracy of traditional algorithms in low cost inertial element attitude determination, this study proposes an improved attitude algorithm based on geomagnetic correction robust model assisted adaptive step gradient descent. Firstly, a robust ellipsoid fitting algorithm is proposed to correct the geomagnetic anomaly data. Then, an adaptive step size model is established based on the momentum gradient descent method, and the iterative optimization parameters are automatically adjusted according to the error function. At the same time, the dynamic limiting mean filter is used to constrain the attitude angle, to calculate the optimal attitude information of the sensor. Finally, the algorithm is verified by static and dynamic experimental data. The results show that the improved algorithm can effectively suppress the angle oscillation, optimize the attitude convergence effect, and improve the accuracy of low cost inertial sensor attitude calculation.