| Title |
Machine Learning Based Optimal Design of a 6.6kW Planar Transformer for High Power Density On-Board Chargers |
| Authors |
Eun-Chong Noh ; Sumi Park ; Gildong Kim ; Jeongwoo Son ; Seung-Hwan Lee |
| DOI |
https://doi.org/10.6113/TKPE.2024.29.1.24 |
| Keywords |
Multi-objective optimization; Planar transformer; Machine learning; On-board charger |
| Abstract |
This paper proposes an optimal design process for a planar transformer for on-board charger(OBC). As it is difficult to mathematically model the high-frequency copper loss of the transformer, conventional design processes tend to have large errors in designing the transformers. In this paper, geometric parameters of a shell-type planar transformer are identified, and FEA simulations are completed to obtain the copper loss of various shapes of the shell-type planar transformers. Regression models for the copper loss are determined using the simulation results and the machine learning technique. From the regression model, optimal high-frequency transformer designs and Pareto front(in terms of transformer 2-dimensional area and loss) are obtained using NSGA-II. In the Pareto front, a design was selected as a desired optimal, and the selected design was verified by FEA simulation. The simulated copper loss of the selected design matched very well, thus demonstrating the validity of the proposed design process. |