Abstract: In the process of geomagnetic field measurement, the three-axis magnetometer is affected by various interference factors and produces various errors which have effect on the measurement accuracy. Based on the analysis of the error sources of measured values of geomagnetic field, an integrated model of three-axis magnetometer measurement is established. An joint estimation iterative algorithm is proposed to calibrate three-axis magnetometer. The performances of the proposed algorithm, extended Kalman filter (EKF) algorithm and nonlinear least squares algorithm are compared. Based on the simulated results, an experimental system was build to verify the proposed method. The simulated and experimental results show that, among the three algorithms, the joint estimation iterative algorithm compensates the error mean and standard deviation to the smallest, and its convergence speed is the fastest. The average error of measured values of three-axis magnetometer can be reduced from 142.4 nT to 12.5 nT, the standard deviation can be decreased from 170.0 nT to 29.2 nT, and the suppression ratio is 84.51%, which provides an effective method for the calibration compensation of three-axis magnetometer. Key

Key words: three-axismagnetometer, jointestimationiterativealgorithm, EKFalgorithm, nonlinearleastsquarealgorithm, nonmagneticrotationequipment