Nonconvex Quadratic Programming Problem for Ship Magnetic Dipole Array Model

The magnetic field modeling methodology for ships based on magnetic dipole arrays demonstrates heightened sensitivity to input data. When addressing overdetermined systems characterized by numerous variables and constrained measurement points, the coefficient matrix frequently develops pathological ill-conditioning, leading to solution divergence and compromised result accuracy. This research reformulates the ship magnetic field inversion challenge as a non-convex quadratic programming problem, employing the Successive Convex Approximation (SCA) algorithm as the computational solver. Rigorous comparative validation was performed against conventional stepwise regression algorithms and experimental datasets acquired from scaled ship model measurements. Results substantiate that while the modeling precision of the SCA algorithm remains comparable to that achieved by stepwise regression methods, SCA exhibits demonstrably superior solution stability. This enhanced robustness positions SCA as a more reliable computational framework for critical naval applications, particularly in high-fidelity ship magnetic signature modeling and rapid detection/localization of magnetic targets in marine environments.