A hybrid domain decomposition and optimization method for predicting electromagnetic emissions from printed circuit boards

Abstract

Electromagnetic emission characterization of printed circuit boards (PCBs) based on phaseless near-field measurement is increasingly concerned. Generally, the PCB is modelled by equivalent elemental dipoles, which are retrieved by optimization method. To accurately model the PCB, enough dipoles are required and consequently an inevitable trade-off occurs between the modelling accuracy and the model retrieval efficiency. To deal with this issue, an efficient hybrid method is here proposed for modelling the electromagnetic emissions from PCBs with equivalent magnetic dipoles. Hybrid domain decomposition and repeated particle swarm optimization method is used to retrieve the equivalent dipoles by fitting them to the amplitude-only magnetic near fields. Prediction of electromagnetic emissions from the PCBs can then be achieved once the equivalent dipole model is determined. The proposed method shows good performance in terms of convergence accuracy and convergence speed. Numerical results are presented to validate the effectiveness of the approach and its robustness against noisy magnetic near fields.