A Triaxial Accelerometer Calibration Method Based on Ellipsoid Hypothesis

刘仁浩,王华,夏金秀

DOI: https://doi.org/10.13494/j.npe.2012.093

2012-01-01

Nanotechnology and Precision Engineering

Abstract:To overcome the problems of high cost,low accuracy and insufficient on-site calibration ability from the existing triaxial accelerometer calibration methods for attitude heading reference system(AHRS),a new method based on ellipsoid hypothesis was proposed for on-site use.Firstly,according to the effect of the sensitivity,bias and non-orthogonal interference on the output of accelerometers,an ellipsoid model for triaxial accelerometer calibration was established,which converts the calibration process of triaxial accelerometer to geometric transformation process from ellipsoid to sphere.Based on the model,the number of calibration parameters was reduced from 15 to 9,and the problem of identifying some non-orthogonal parameters was solved.Secondly,the likelihood estimation objective function with variable weights was established.According to the errors of each accelerometer′s measuring data,the corresponding weights were assigned to these error values,i.e.minor weights were assigned to the measuring data with major error,thus to weaken the effects of sensor noise.Finally,the maximum likelihood estimation calibration parameters were calculated through particle swarm optimization(PSO) algorithm,and the calibration method was used to calibrate the triaxial accelerometer of an AHRS.Experimental results show that the roll angle error is 5 times lower than the one before calibration using the proposed method without auxiliary equipment,which can satisfy the requirement of the AHRS.