Magnetic Field Calculation with Generative AI

This study presents a novel approach to overcome the limitations of traditional Finite Element Analysis (FEA) for magnetic components, which are time-consuming and require specialized knowledge. By using neural networks, the simulation process is simplified. Thousands of 2D and 3D simulations were conducted, and the neural network design, comprising an initial Neural Network (NN), and a following pix2pix network, effectively predicts and refines the magnetic field. This method reduces computational time from seconds to milliseconds and achieves deviations of less than 5 %.