A Phaseless Source Reconstruction Method Based on Hybrid Dynamic Differential Evolution With Least Square and Regularization

In this article, we present a phaseless source reconstruction method based on dynamic differential evolution (DDE) combined with the least square and regularization for effectively constructing the equivalent dipoles to represent the real electromagnetic interference (EMI) sources, which enables the accurate reconstruction of near fields. The conventional DDE method is sensitive to the initial values of dipole moments, and the improper initialization may lead to the failure of its convergence. To avoid this, the proposed method employs least square (LSQ) to predict the dipole moment levels during initialization. This eliminates the need of multiple attempts to determine the appropriate range of dipole moments and thus saves the computation time. Moreover, LSQ is embedded in each iteration of DDE to further manipulate the dipole moments. The effectiveness and accuracy of the proposed method are verified through both simulation and experiment. This article offers a new and more reliable alternative for accurately reconstructing the near fields of EMI sources.