To solve the problems existing in three-axis magnetometer error correction method as such amount of computation is large

reliance on outside reference information is excessive

and online calibration is unrealizable

an online self-calibration method based on recursive least square is proposed. According to description by the Poisson equation on measurement model

magnetic field vector error correction model is derived. Rrelationship between calibration model and coefficients of ellipsoid surface equation is established based on theory of ellipsoid hypothesis. Realization process of on-line identification of coefficients of ellipsoid equation is derived based on recursive least squares

and then obtain error correction parameters. The results show that this method can correct effectively data distortion caused by soft and hard magnetic effects. And the maximum fluctuation range of magnetic field intensity of sampling points is reduced from 67. 112 8 μT to 14. 064 8 μT

and the error standard deviation is reduced from 15. 828 7 μT to 6. 345 1 μT

which indicate that the proposed method is suitable for error selfcalibration of three-axis magnetometer without external reference Benchmark.