Parallel Frameworks for Robust Optimization of Medium-Frequency Transformers

Kristen Booth,Harish Subramanyan,Jun Liu,Srdjan M. Lukic

DOI: https://doi.org/10.1109/jestpe.2020.3042527

IF: 5.462

2021-08-01

IEEE Journal of Emerging and Selected Topics in Power Electronics

Abstract:Current optimization methods for medium-frequency transformers (MFTs) within power electronic converters yield unrealistic results in the multiphysics framework. Comparing the optimal design to an experimental setup for a 3.5-kW MFT, the core loss is underestimated by 28%, which results in the experimental steady-state temperatures being 10 °C greater than the analytically optimized model. To counteract these disadvantages, an optimization procedure, using the aggressive space mapping (ASM) technique, is experimentally verified and compared with the previous state-of-the-art (SOA) method. It is shown that the ASM design produces more realistic and feasible experimental outcomes than the SOA design. The core losses are accurately predicted to within 10%, which, in turn, vastly improves the thermal modeling accuracy. The ASM method accurately predicts the core hot spot temperature and the average core temperature. This work also introduces a robust optimization method to the MFT design process to handle variations from both converter-level attributes and manufacturing tolerances to create a potential design region, which contains 97.725% of possible design outcomes. This method replaces the nominal design optimization that is used to produce the optimized MFTs in the SOA and ASM methods.

engineering, electrical & electronic