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.

中文翻译：

---

中频变压器稳健优化的并行框架

电力电子转换器中中频变压器 (MFT) 的当前优化方法在多物理场框架中产生了不切实际的结果。将优化设计与 3.5 kW MFT 的实验设置进行比较，铁损被低估了 28%，这导致实验稳态温度比分析优化模型高 10 °C。为了抵消这些缺点，使用激进空间映射 (ASM) 技术的优化程序经过实验验证，并与之前的最先进 (SOA) 方法进行了比较。结果表明，ASM 设计比 SOA 设计产生更现实和可行的实验结果。磁芯损耗准确预测在 10% 以内，这反过来又大大提高了热建模精度。ASM 方法可以准确预测核心热点温度和平均核心温度。这项工作还为 MFT 设计过程引入了一种稳健的优化方法，以处理转换器级属性和制造公差的变化，以创建一个潜在的设计区域，其中包含 97.725% 的可能设计结果。此方法取代了在 SOA 和 ASM 方法中用于生成优化 MFT 的标称设计优化。