Research on magnetic field inversion and magnetic field evaluation algorithm based on minimum mean square error

In order to ensure the stability of the inversion model，the mathematical model of magnetic field inversion modeling was improved based on the point magnetic charge theory．Aiming at the problem of poor fitting effect caused by small y component of magnetic field in fixed north-south course measurement，an inversion algorithm based on minimum root-mean-square error (MMSE) with adaptive weights was proposed，which was verified by measured magnetic field data to ensure that all three components of magnetic field could achieve relatively high precision depth conversion．After adding 10 dB noise，the error is still less than 10%，which accords with the engineering practice，and proves the stability of the model algorithm．At the same time，in order to evaluate the correctness of inversion input magnetic field data，a magnetic field evaluation method based on T-distribution principle was proposed．The results show that the magnetic field confidence region can be obtained by this method，and the magnetic field data meeting the requirements in the original sample can be screened with 95% confidence，and the screening efficiency is 20%，which improves the probability of the correctness of the inversion input data and realizes the effective evaluation of multiple groups of magnetic field data. © 2023 Huazhong University of Science and Technology. All rights reserved.