Modeling and Optimizing Winding Arrangement for Gapped Planar Magnetics based on Artificial Neural Network

Fringing loss dominated nonlinear winding loss in gapped inductor poses a great challenge for magnetic loss characterization due to the lack of full physical models. This paper investigates the impact of winding arrangements on the AC resistance in gapped inductors, using an EI core as a case study. It specifically examines how the distance of the conductor from the gap affects AC resistance. The study reveals that asymmetric winding arrangements exhibit lower AC resistance compared to symmetric ones. To facilitate rapid computation of eddy current losses, the generation of frequency-based AC resistance maps, and the analysis of AC resistance under various winding configurations, a software tool based on an artificial neural network (ANN) model is developed. The proposed ANN-based model uses 5.7 μs with a deviation less than 4.29% with respect to FEA simulations.