Calculation of Iron Loss in Soft Magnetic Composites Using Neural Network-Based Dynamic Hysteresis Model Under SVPWM Excitation

• Published in: 2024 IEEE 21st Biennial Conference on Electromagnetic Field Computation (CEFC) pp. 1 - 2
• Main Authors: Jing, Ying, Zhang, Yanli, Zhang, Dianhai, Zhu, Jianguo
• Format: Conference Proceeding
• Language: English
• Published: IEEE 02.06.2024
• Subjects: Artificial neural networks; Computational modeling; Dynamic hysteresis models; Iron; Magnetic cores; Magnetic hysteresis; Motors; Neural network; Soft Magnetic Composites; Space vector pulse width modulation; Space Vector Pulse Width Modulation excitation
• DOI: 10.1109/CEFC61729.2024.10585744

Soft Magnetic Composite (SMC) materials, serving as key components in high-power motors, exhibit a loss mechanism distinctly different from traditional laminated core materials like silicon steel. To predict these losses more accurately, this study introduces a dynamic hysteresis model based on Neural Networks (NN) and trapezoidal equivalent circuits, designed to calculate the iron losses of SMC under Space Vector Pulse Width Modulation (SVPWM) excitation. A NN algorithm, capable of characterizing the magnetization process, simulates the static hysteresis component. Meanwhile, when calculating eddy current losses and anomalous losses using trapezoidal equivalent circuits, the effect of skin effect on iron losses is taken into account. The model's accuracy and effectiveness are validated through a comparison of theoretical and experimental results. This model enhances the application of SMC in motors and provides robust support for the optimization of motor design.