Conference Agenda

This paper proposes a surrogate model for the rapid evaluation of electric machine designs, based on a neural network combined with a semi-analytical subdomain model. Although both analytical physical-model approaches and data-driven approaches have been proposed to construct surrogate models, which can be significantly faster than numerical finite-element simulations, issues still remain. On one hand, simplifications in analytical approaches often cause inaccuracy, especially in the prediction of highly nonlinear phenomena such as cogging torque of permanent magnet synchronous motors; on the other hand, purely data-driven approaches often require a large amount of training data to achieve high accuracy. In our proposed method, the performance of the electric machine is initially approximated by using a semi-analytical subdomain method, and this initial prediction is used as the input of a neural network, together with other design variables, to obtain the final prediction. We test the method to predict the cogging torque of surface-mounted permanent magnet motors. By combining physical-model and data-driven approaches, the proposed method can predict cogging torque with good accuracy, which cannot be achieved with only physical-model; the prediction accuracy is also much improved compared with conventional neural networks, especially when the size of the training dataset is small.

This paper presents a method to take into account, in a circuit model, the frequency-dependent behavior of electrical machine windings. First, the frequency-dependent inductances (both self and mutual) and resistances are determined thanks to time harmonic finite element (FE) computations in the frequency range of interest. The frequency-dependent behavior of the capacitances is investigated and the choice of using an electrostatic model is justified. Then, the different frequency-dependent parameters are included in a circuit model using behavioral voltage sources placed in series, which allows to forego delicate fitting techniques. A test winding with 69 turns is used as a validation of the proposed method. An excellent agreement between experimental and simulation results is observed in the frequency domain. Time-domain node voltages can be computed by an Inverse Fast Fourier Transform after reducing the circuit to the nodes of interest, which enables the usage of the proposed method as a diagnostic tool before or during the design phase electrical machine winding design.

This article proposes a Cauer Ladder Network (CLN) with constant basis functions for eddy current magnetic field problems involving translational movement. The analysis domain is decomposed into two domains: the stator part and the mover part. The CLN method is applied to each domain to derive circuit parameters and basis functions corresponding to each CLN. Interactions between two domains are modeled by the current-controlled voltage sources connected to each stage of the CLN. The gains of the current-controlled voltage sources are computed by the Galilean transformation Biot-Savart law. To show the validity of this proposed method, the analysis of the TEAM Workshop Problem 28 was performed, and the obtained steady-state eddy current field distributions were compared to that obtained by the conventional Finite Element Method (FEM).

In this article we propose an efficient calculation method for the estimation of in-plane eddy currents generated in large induction motors that consists of core packets. In the calculation, the 3D analysis region is axially separated into one end core packet and half the inside packet by assuming an approximation of the boundary condition between the core packets. By using this method, the losses at each electrical steel sheet can be obtained with practical computer resources within a short computational time. After confirming the validity of the approximation, the proposed method is applied to estimate the loss at each electrical steel sheet. It is clarified that the core loss including the in-plane eddy current loss is highly concentrated at one electrical steel sheet at the end.

In order to increase the speed range of permanent magnet machines in constant power regions, different techniques including multi-phase configurations have been adopted. This paper presents the dual three-phase operation of a novel dual-airgap permanent magnet vernier machine having a yokeless rotor for an extended speed application. The presented machine contains two stators each having a three-phase winding, and a single yokeless rotor. The outer and inner stator windings are treated as two different three-phase windings to apply the six-phase or dual three-phase operation. The winding currents are either out-of-phase or in-phase based on the current angle between two windings. The 2D-FEM analysis is used to analyze the performance of the presented machine for different combinations of dual-three-phase operations. The detailed results and discussion will be included in the full version of this manuscript.

This paper is mainly concerned with the parasitic parameters identification of the high-frequency transformers, namely the parasitic capacitance and the leakage inductance. With the increase of the operating frequency of the power electronic switches, the parasitic effects are getting more and more important as well as the displacement current which is unreasonable to be neglected in the high-frequency range. Unlike the traditional way to identify these parameters by solving the electro-quasistatic and the magneto-quasistatic Maxwell models respectively, the Darwin model in the frequency domain is employed to study the capacitive and the inductive effects simultaneously. The inductance and capacitance vary with the frequency of a 2-dimensional high-frequency transformer (HFT) with foil windings will be studied by the Darwin model, and the result will be compared with those solved by the electro-quasistatic and the magneto-quasistatic models and validated by the measurement data. Finally, it is believed that the parasitic parameters extracted by using the Darwin model can give more accurate results for HFTs operating in high-frequency conditions since the interaction of both inductive and capacitive effects is calculated simultaneously. It is helpful for the design and better understanding of the performance analysis of HFTs.

In the nonlinear magnetic field analysis of an inductor driven by inverter power supply, the number of time steps becomes huge to take account of harmonics. To reduce the computation cost, the fixed-point harmonic-balance (FPHB) method is examined in this paper. The FPHB method is based on the frequency domain and can solve the nonlinear magnetic problems. It is applied to simplified nonlinear 1D inductor models with the current and voltage sources including harmonics. It is shown that the computation cost can be reduced by focusing on the dominant frequency components of the flux densities when the voltage source is applied.

When the permanent magnet synchronous motor (PMSM) is manufactured, the magnetization process of the permanent magnet (PM) is carried out post assembly. After the post-assembly magnetization process, the estimation of the magnetization quality is essential for designing a high-performance IPMSM. There are two types of the magnetization estimation methods for the permanent magnet implemented on the magnetic circuit: the first is estimation using a magnetic sheet, and the second is based on the magnetic flux measurement using a Gauss meter outside the magnetic circuit. Although these methods simply estimate the magnetization quality of PM, the detailed magnetization information is not obtained. In this paper, to enhance the design of a high-performance PMSM, the Magnetization Estimation method using the Leakage Flux of a magnetic circuit with magnetic nonlinearity (MELF) is proposed, and its performance is verified using the fundamental magnetic circuit.

In this paper, a new analytical model for predicting the magnetic field of the U-shaped interior permanent magnet motor (IPMM) is proposed. In the polar coordinates, rectangular permanent magnets are equivalent to the combination of consistent fan-shaped subdomains, and after that, the Laplace or Poisson equation of each solution domain can be obtained. Furthermore, based on the harmonic modeling (HM) technique, the saturation characteristics of the magnetic bridge are accurately considered, and then its magnetic field distribution (MFD) and electromagnetic performances (EPs) can be obtained. The approach proposed in this paper includes but is not limited to U-shaped IPMMs, and it can also be applied to spoke-type or V-shaped IPMMs with a minor modification.

In the ring core inductor driven by the inverter using GaN devices, the ringing phenomenon occurs in the generated current due to the short rising and falling time of the input voltage. This ringing phenomenon is due to the resonance at resonant frequencies in the ring core inductor. To investigate the mechanism of resonance at high frequency (HF) in detail, the HF characteristics of ring core inductor are analyzed in frequency domain using the electromagnetic field analysis taking account of the displacement current. It is shown that the resonance at some frequencies can be simulated, and it can be explained as the phenomenon of the wave propagation in the distributed constant circuit.

Unequal tooth is useful structure to enhance flux linkage of the motor with single-layer concentrated winding.

Finite-element method(FEM) is required to accurately predict the electromagnetic field distribution of the motor. However, it is computationally expensive, especially in the early design stage.

In this paper, we propose a subdomain method for predicting magnetic field of surface-monted permanent-magnet synchronous motor(SPMSM) with unequal tooth.

With the proposed method, the air-gap flux density distribution of SPMSM with unequal tooth structure were precisely calculateed. The proposed method was verified by FEM.

Understanding physical effects occurring for example in the electromagnetic field can be a challenging task. In order to ease the learning experience it is beneficial to visualize and encourage interaction with the physical field. Augmented Reality (AR) can serve as a tool to visualize naturally invisible fields in order to help students to understand physical effects. In this paper, we present a workflow on how to incorporate field results stemming from a FEM-tool or simple analytical solutions into an augmented reality (AR) experience. The focus hereby lies on providing a simple framework that can be used by educators to incorporate this tool into school or university teaching. We present a workflow on how to process simulation results so that they can be used with AR and provide source material by means of a template and a guide so that educators can quickly translate their projects into (augmented) reality. The basis for this project is the game engine Unity which can be used free of charge for educators. Combined with other free or open source programs for visualization and preparation like openCFS and Paraview, this setup can be used freely by anybody for means of education.

This paper proposes a magnetic field visualization system for Mixed Reality (MR). We have proposed a visualization system for education using AR and VR. As a further development of this, we are applying a visualization system for education using MR technology. We took a Hololens2 as an MR device. A system for MR allows us to observe a virtual world while having a vision of the real world. Because of this characteristic, a visualization system for MR can draw analysis results at any place, on or beside an actual analysis target. Our proposed system allows users to model an analysis target consisting of voxels with their fingers. The system can also draw a magnetic field in the form of a vector field. Visualizing the vectors not at every voxel but only on an optional plane allows users to see the magnetic field concisely. The system can draw multiple models at any place regardless of the position of markers like AR.

Edge Elements (EE) are widely used since the 1990s. In particular, first order elements (tetrahedron for 3D cases) are the most

employed because they are relatively simple and well adapted to complex geometric domains. Several codes propose meshing for such

elements. Nevertheless, the understanding of the corresponding shape functions is not easy and often their well know expressions are just

used without an appropriate presentation. The aim of this paper is to propose a rationale for these shape functions showing a relatively

smooth and understandable explanation. In this paper, we use a 2D first order triangle (which has similarities to the 3D tetrahedron) for

a direct and simple algebraic presentation.