Conference Agenda

The magnetic characteristics of the oriented silicon steel used in transformer cores vary with material properties, external frequency, and mechanical stress. Adverse operating conditions can exacerbate the losses in oriented silicon steel sheets and enlarge the magnetostrictive strain, leading to noise issues. In this paper we present an enhanced loss separation model and a magnetostrictive model for oriented silicon steel sheets under complex operating conditions. These two models are established on the basis of Bertotti's traditional loss separation model and Sablik-Jiles' magnetostrictive model. Drawing upon ferromagnetism theories, such as excess loss, hysteresis loss, and stress-induced magnetostrictive strain, a series of experimental investigations have been carried out on the loss and magnetostriction characteristics of various models of oriented silicon steel sheets under varying stresses and frequencies. These experiments validate the effectiveness and accuracy of the enhanced loss separation model and magnetostriction model. Furthermore, by utilizing the enhanced model proposed in this article, the loss characteristics and magnetostriction variation of oriented silicon steel sheets under complex operating conditions are examined, thereby confirming the precision of the two models. The findings presented in this paper serve as a crucial theoretical foundation for the reduction of loss and noise in transformers.

This paper investigates the influence of B-H extrapolation methods on the transformer characteristics under GIC conditions. Four commonly used B-H extrapolation methods are applied, and their effects on the magnetic properties of grain-oriented electrical steel sheets in the saturation region are explained. Considering these magnetic properties, the magnetic and thermal characteristics of the transformer are evaluated using 3D finite element method. Through numerical estimations applying various extrapolation methods, this study demonstrates that the choice of B-H curve extrapolation method is a crucial factor in predicting transformer characteristics under GIC conditions.

This study proposes a rapid prediction method for transformer noise considering material parameter uncertainty. To precisely characterize the uncertainty in silicon steel material parameters, the magnetization and magnetostriction laws are randomly modeled using copula theory, generating correlated sample curves. Then, to enhance the prediction method for transformer noise responses, an adaptive differential evolution algorithm-optimized backpropagation neural network (ADE-BP) is proposed. Finally, the proposed method is applied to the magnetic-structural-acoustic multi-field coupling model of the transformer. The simulation results illustrate the effectiveness and accuracy of the proposed approach.

Multi-scale energy models (MSEM) are commonly utilized in the magnetic properties modeling of polycrystalline materials. To improve the computational efficiency of MSEM, this paper investigates a parallel computing method to obtain the demagnetization coefficient and to carry out the minimization of magnetic domain energy. In this study, a grain-oriented electrical steel is discretized into many single crystal elements in accordance with the actual size of the sample. The magnetic properties of electrical steel sheets under different numbers of discrete elements are explored. The computing results indicate that the proposed model can improve computational efficiency as well as good agreement with the experimentally measured values.

In this paper, a generalized inverse Preisach model based on neural networks is proposed by characterizing an Everett function of the irreversible magnetization and taking magnetization-dependent hysteresis, as well as reversible magnetization part, into account, which not only ensures the simulation accuracy for both symmetric and asymmetric minor loops but also very easy to implement the model in the FEM. The model accuracy is verified by the experimental data of B30P105 grain oriented (GO) silicon steel under different magnetization conditions, and the computational efficiency is compared with the traditional model.

We propose a novel approach for modeling soft magnetic materials (SMMs) at high frequency under both dielectric and magnetic excitations in time-domain. Cauer networks (CNs) are first synthesized from analytical material models obtained by homogenization for modeling the effective electromagnetic behavior of SMMs in both frequency- and time-domains. The network components depend explicitly on the physical and geometric properties of the material microstructure. The CNs are then coupled with finite element (FE) models to simulate the macroscopic behavior of SMMs in both frequency- and time-domains. Finally, numerical simulations are validated against measurements. Mn-Zn ferrite and soft magnetic composite samples are investigated. The proposed approach provides a practical framework for efficiently modeling SMMs under different excitations in both frequency- and time-domains.

Accurate hysteresis modeling with minimal use of experimental data sets is essential for the design and performance optimization of highly efficient magnetic materials. However, the accuracy of the traditional Preisach model is often inadequate, especially under complex excitation conditions. To solve this challenge, a Preisach operator intelligent hysteresis (POIH) model is proposed in this paper. By combining Preisach operator with backpropagation neural network, the model combines the simplicity of traditional physical model with the high precision and flexibility of modern data-driven model. Based on the experimental data of 30Q130 silicon steel, the model is trained and tested under the conditions of stress and temperature. The experimental results show that the model has advantages in accuracy, efficiency, and universality.

This work focuses on the vectorization of the magnetic anisotropy model for grain-oriented steel sheets by exploiting coenergy computation and interpolation methods. The proposed approach is based on a modified representation of coenergy isocontours in elliptical form and introduces further improvements with the integral form of the error functions and Fourier series expansion. This will provide a remedy for the deficiencies of the aforementioned approaches and introduce a more robust framework for modeling and simulating the anisotropic magnetic response.

This paper presents a vector hysteresis model that characterizes the anisotropic behavior of grain-oriented (GO) electrical steel. The anisotropic anhysteresis model with permeability tensors that considers crystallographic orientation based on magnetic domain theory is developed. An additional easy axis aligned with the direction of the applied magnetic field is introduced for anisotropy near saturation. The hysteresis behavior is modeled through the association of the anhysteresis model and energy-based model. Material parameters for both reversible and irreversible behaviors are identified through experimental characterization performed on B27R090 electrical steel under different magnetization directions. Comparisons between experimental and calculated hysteresis loops, as well as iron losses, validate the model's accuracy.

Post-assembly magnetization (PAM) simplifies the manufacturing of permanent magnet electrical machines, but faces challenges in large size machines due to the limited magnetizer power. This paper proposes a PAM method with induction heating to reduce the required magnetizing field intensity, utilizing the temperature-dependent property of the Nd-Fe-B magnets. The same coils are used for both heating and magnetization by adjusting the magnetizer parameters. The method can achieve saturated magnetization in large machines with limited magnetizer power,

demonstrating its practicality and effectiveness.