Bi-objective optimization of magnetic core loss and magnetic energy transfer of magnetic element based on a hybrid model integrating GAN and NSGA-II

Magnetic core loss and magnetic energy transfer are important indicators of the performance of magnetic elements in electronic devices. However, there is a restrictive relation between those two indicators. For example, increasing magnetic field intensity or frequency will enhance the transmission capability of magnetic elements but will exacerbate magnetic core loss. Conversely, reducing the magnetic field intensity or frequency will decrease magnetic core loss but compromise transmission efficiency. To address this challenge, this study proposes a bi-objective optimization method for magnetic elements under complex operating conditions (such as high-frequency and non-sinusoidal waveforms), which integrates Generative Adversarial Network (GAN) and Non-dominated Sorting Genetic Algorithm II (NSGA-II). The role of the GAN module in the hybrid model is to generate a diverse initial population to expand the search space and improve population diversity. Meanwhile, the role of the NSGA-II module is to apply non-dominated sorting and crowding distance calculations to optimize magnetic core loss and magnetic energy transfer. Experimental results show that the proposed optimization model effectively reduces magnetic core loss and significantly enhances magnetic energy transfer under high-frequency and non-sinusoidal conditions. Furthermore, a comparison with the NSGA-II algorithm is conducted to verify the efficiency of the bi-objective optimization model. • A bi-objective optimization method combining GAN and NSGA-II is proposed. • The hybrid optimization model reduces magnetic core loss and enhances magnetic energy transfer. • Analysis of variance and Spearman correlation are used to identify key factors.