Multi‐Objective Robust Design of High‐Frequency Transformers Considering Material Properties and Process Deviations

ABSTRACT

The design of high‐frequency transformers (HFTs) aims to maximise power density and efficiency. However, the variations in the manufacturing process from the raw material to core result in nonuniform magnetic properties. Although empirical correction factors are usually used to address such uncertainties, significant discrepancies remain between predicted and actual performances. This paper presents a robust design method for HFTs based on Design for Six‐Sigma (DFSS). The uncertainties such as the loss parameter deviations of the nanocrystalline core and Litz wire and the insulation distance tolerances are incorporated into the optimization framework. The modified nondominated sorting genetic algorithm II (NSGA‐II) is presented, followed by a comparative study of five 20‐kW, 10‐kHz prototypes designed using the conventional deterministic design and the proposed robust design method. The experimental results show that the prototypes designed by the proposed method satisfy all specifications and exhibit reduced performance deviations, thus verifying the effectiveness and engineering practicality of the proposed method.