Analyses of Triaxial Accelerometer Calibration Algorithms

Abstract

This paper proposes a calibration procedure in order to minimize the process time and cost. It relies on the suggestion of optimal positions, in which the calibration procedure takes place, and on position number optimization. Furthermore, this paper describes and compares three useful calibration algorithms applicable on triaxial accelerometer to determine its mathematical error model without a need to use an expensive and precise calibration means, which is commonly required. The sensor error model (SEM) of triaxial accelerometer consists of three scale-factor errors, three nonorthogonality angles, and three offsets. For purposes of calibration, two algorithms were tested-the Levenberg-Marquardt and the Thin-Shell algorithm. Both were then related to algorithm based on Matlab fminunc function to analyze their efficiency and results. The proposed calibration procedure and applied algorithms were experimentally verified on accelerometers available on market. We performed various analyses of proposed procedure and proved its capability to estimate the parameters of SEM without a need of precise calibration means, with minimum number of iteration, both saving time, workload, and costs.