A method for gravitational apparent acceleration identification and accelerometer bias estimation

Abstract

Accelerometer bias in an inertial navigation system (INS) is the key factor determining the navigation accuracy. Two methods, named separated calibration and system calibration are always used for acceleration bias calibration. The former one needs a high precision turntable to provide reference and the latter cannot estimate accelerometer bias along horizontal direction because of coupling effects between horizontal misalignment angles and horizontal accelerometer bias. In this paper, a new estimation method is designed to estimate accelerometer biases and acquire high horizontal alignment accuracy without a turntable. In the proposed method, different expressions about gravity and accelerometer biases in an inertial frame are analyzed and coupling models about gravity and biases are constructed. Then, measurements from gyroscopes and accelerometers are used to construct gravitational apparent acceleration and Kalman filter used to estimate parameters describing apparent acceleration and estimate accelerometer bias. The simulation and turntable tests show that when the carrier is without translational but with swinging motion, the proposed method can finish gravitational apparent acceleration identification and accelerometer bias estimation effectively at the same time, and when the reconstructed gravitational apparent acceleration used for alignment, horizontal alignment errors can nearly approach zeros.