Aeromagnetic Compensation Based on Truncated Singular Value Decomposition With an Improved Parameter-choice Algorithm

As aircraft magnetic interference is much larger than the resolution of magnetic sensors, aeromagnetic compensation is increasingly important with the development of modern airborne magnetometers. Due to the ill-conditioning of Tolles-Lawson model, some regularization method, instead of ordinary least-squares (OLS), has to be employed to predict the magnetic interference caused by aircraft's motion. This paper aims to provide a candidate method to reduce such interference. The truncated singular value decomposition (TSVD) is used as a main tool to estimate compensation coefficients. In order to select an appropriate regularization parameter for TSVD, an improved parameter-choice method, L-curve criterion with defined origin algorithms (LO) is proposed in this paper to locate the global corner of the discrete L-curve rather than small local ones. Simulation results demonstrate that the proposed combined method (TSVD-LO) can reduce the magnetic interference to a large extent.