Cooperative Suppression of Negative Effects Associated With Multicollinearity and Abnormal Data for Aeromagnetic Compensation

Compensating the magnetic interference related to airplane maneuvers is essential to high-precision aeromagnetic measurement. In estimating the compensation coefficients, aeromagnetic compensation based on the Tolles-Lawson model needs to address the reduction in accuracy and robustness arising from several negative effects. Because the existing methods only suppress single effects independently, the scope of their application is very limited. To address this problem, a cooperative suppression method of the negative effects associated with multicollinearity and abnormal data is proposed in this study. For this method, abnormal data produced by the dead-zone effect of a classical optically pumped magnetic sensor are assigned lower weights, thereby reducing their negative influence in the estimation of the compensation coefficient. Moreover, a ridge parameter is added to further reduce the multicollinearity of the compensation coefficient. Theoretical evaluation shows that in the presence of 20% abnormal data and 0.5 correlation, this method reduces the mean square error (MSE) for the coefficient to 0.76 compared with 4.39 for the least-squares (LS) method, 3.43 for the ridge method, and 2.13 for the robust method. For practicality verification, we built an experimental platform and mounted it on an aeromagnetic survey airplane, and then performed a compensation flight test. The results demonstrate an improved ratio for our method of 12.79, which is significantly higher than 0.56 for the LS method, 2.73 for the ridge method, 4.29 for the robust method, and 7.95 for the state-of-the-art commercial compensator.