Conference Agenda

In this paper we propose a method to provide a wideband 3D model of a toroidal shaped common mode choke (CMC). The overlapping turns of the CMC’s coils are treated by equidistant turns and lumped element external capacitors, while the complex, frequency dependent material parameters—permittivity and permeability—are modeled by a 2nd order Debye model. Obtaining the optimal capacitance values and Debye model parameters is addressed as an inverse problem that leads to a minimization of the error between the simulated—with 3D finite element (FE) method—and measured transfer impedance curves of the CMC. The computational cost is reduced by a polynomial chaos kriging surrogate model that is built using a small sampling design and is subsequently used as a low-cost substitute of the FE model to support the exhaustive search that the inverse problem implies.

This paper presents a design methodology for electric machines (EM). It is based on discrete multi-material topology optimization (TO) and printed material characteristics. Our methodology uses the Bi-directional Evolutionary Structural Optimization (BESO) heuristic for optimization, and to maximize EM torque, we propose a reformulated problem to accelerate simulation time. It permits considering both ferromagnetic materials and permanent magnets with discrete variables. Printed material characteristics are taken into account, considering several process parameters. A comparison of the final designs for different BH curves is presented

Featuring a binary component micro-structure, Soft Magnetic Composites (SMCs) offer numerous benefits. Topology optimization (TO) is an effective method of redistributing materials to achieve optimal physical properties. This research aims to develop an optimal microstructure of SMCs based on a TO algorithm using density-based design methods. Regulation and projection are adopted to assure convergence. The objective of the optimization is the effective permeability of the composite in the frequency range, which is implicitly carried out through magnetic energy. Rapid convergence is guaranteed due to the gradient-based algorithm. Typical parameters for SMCs applications are calculated and optimized. The optimal structure of SMCs is consistent with our analytical study.

This paper presents a fast optimization method of multi-turn tubular coils for high-frequency wireless power transfer system. In this method, the homogenization method to quickly estimate the AC losses of the wireless coils is employed. The optimization for high-frequency wireless coils using tubular wire-formed coil is performed to enhance the transfer efficiency. It is shown that the proposed method can dramatically reduce the optimization time comparing to that by the conventional optimization method.

In this paper, we propose a design method with Autoencoder (AE) to learn the correlation between objective function values and design shapes and efficiently extract shape information that improves the objective function values in design optimization for complex actual models by dimensionality reducing the design space. The novelty of the proposed method is that the network parameters are learned so that some latent variable components represent highly correlated features with specific objective functions. As a result, it is possible to output shape changes corresponding to increases or decreases in multiple objective function values independently and continuously, which provides designers with useful design strategies. In addition, by using the generated shape as the initial one for level-set optimization, we can derive a new structure with manufacturability not found in the training data, which results in an efficient global solution search.

This study reports on a design method for thin dielectric lenses for millimeter-wave antennas using high dielectric constant materials. In 76 GHz band millimeter-wave regions used for automobile radar and other applications, because the antennas become smaller, the dielectric lenses loaded on the antennas are also expected to become smaller and thinner. The purpose of this study is to develop thin dielectric lens by utilizing the wavelength shortening effect of high dielectric constant materials. In order to design thin dielectric lens, the topology optimization using the normalized Gaussian network (NGnet) is introduced. This paper presents an example of thin lens design using high dielectric constant material when the relative permittivity is 10. In addition, design example of thin lens using gradient functional materials with different relative permittivities are shown for comparison.

In this study, we propose an optimal design method to reduce current ripple through optimization with maximize motor inductance and to improve the design speed and accuracy through PWM harmonics injected current source. Design with PWM current source analysis is ideal because it is possible to precisely analyze the iron loss and torque ripple characteristics of electric vehicle (EV) traction motors affected by current ripple. However, the application of PWM current source in the design of EV traction motors is limited. Because it has design difficulty according to nonlinearity characteristics and complexity of design parameters. Various methods are being researched to reduce the time of the design which adopts PWM current source analysis, but it is more fundamental to consider reducing the current ripple. Our proposed method is effective in terms of design speed and accuracy compared to conventional design methods. We reduce the current ripple by maximizing motor inductance that affects the current ripple characteristic within a range that satisfies the target performance in optimization. Then, apply the PWM switching harmonics to the ideal current source to derive results corresponding to the actual. For this, we verify the efficiency of our method by comparing it with the results of PWM current and ideal current source analysis, respectively.

Recently, the need for collaborative robots is increasing as they can replace workers' repetitive manual work to improve work efficiency and solve labor shortage problems. The motor used for the joint of the collaborative robot must satisfy the limited axial length to ensure safety while having a large driving range in a limited space. Therefore, in the size of a thin structure with a short axial length, a design was performed to improve the output by replacing the existing radial permanent magnet motor (RFPM) with an axial flux permanent magnet motor (AFPM). In RFPM, the torque is proportional to the square of the diameter of the motor and the stacking length, so if the axial length is short, the performance is reduced in proportion to it. On the other hand, since the torque density of AFPM is proportional to the cube of the diameter of the motor and the permanent magnet can be used as much as possible as long as the core is not saturated, the output can be higher than that of RFPM in the same volume. In this paper, the design process for selecting the shape of AFPM that can increase the power density of the axial flux motor was proposed and the final model was selected. The combination of the number of pole slots was selected, detailed design for shape variables for various types was carried out, and the feasibility of the study was verified by performing finite element analysis (FEA).

An adaptive compactly support radial basis function (ACS-RBF) based response surface methodology (RSM) is proposed to extract frequency-dependent stray parameters of an advanced power electronics device. Based on a posterior error with the results obtained by conventional complicated simulation methods, the proposed ACS-RBF methodology can automatically select the sampling points and determine the number of sampling points while constructing the surface response model which satisfies the predefined accuracy. The computational efficiency and accuracy of the proposed method are evaluated by the numerical results on solving the transient process of a prototype insulated-gate bipolar transistor (IGBT).

In our previous work, we proposed a Hemocompatibility Assessment Platform (HAP) to quantify blood trauma caused by individual Left Ventricular Assist Device (LVAD) components. We used a combination of Magnetic-Levitation (Maglev) bearings and BLDC motors to support and rotate the rotor impeller without any contact to eliminate mechanical wear and hemolysis caused by the platform. Passive magnetic bearings (PMBs) that utilize the repulsive force of permanent magnets provide radial direction control of the rotor-impeller. The resultant axial instability forces generated by the PMB are supported actively by an Active Magnetic Bearing (AMB) that combines permanent magnets and electromagnets. The suspension force stiffness of the AMB must exceed the negative spring stiffness of the radial PMB to meet the target controllable range. Therefore, a detailed design of the axial AMB was created through magnetic field analysis using the finite element method (FEM) to design an axial AMB that can generate the required support force under the structural size constraints. Based on the design results, an experimental apparatus was fabricated. Then, a maglev motor system for the proposed HAP was developed.

This paper analyzes the sensitivity of a shaft voltage (SV) for a consequent-pole magnet motor (CPPM) considering the magnetic flux path according to the configuration of the magnet and flux barrier. Moreover, we carry out an optimal design to improve the SV and torque characteristics based on the analysis results. First, the design parameters affecting the shaft leakage flux, which causes the SV, are determined. Then, a sensitivity analysis of the influence of the design parameters on the SV is carried out. Finally, we design a CPPM to reduce the SV by applying an optimization algorithm based on the results of the sensitivity analysis.

In this paper, a novel hybrid excited permanent magnet synchronous machine (HEPMSM) is presented which can enhance the capability of flux regulation and improve the flux leakage at the ends of rotor. The presented machine consists of two radial stators, two radial rotors and an axial stator, and has both radial and axial magnetic circuits. The side of the rotor is machined to form several sector rings so that the flux flows into both radial and axial stators. The machine has two magnetic pole structures, NN type and NS type, which will affect the flux path of the machine. This paper designs a machine prototype and gives the structure of the machine. First, analyzes the operating principle of the two magnetic pole structures in detail. Then, the 3-D FEM is established, and the basic electromagnetic performance of the machine is calculated and compared. Lastly the flux regulation capability of the axial stator is analyzed. The analysis shows that the machine has fine flux regulation capability and high power density, therefore, it has advantages in the fields of wide speed range such as electric vehicles.

A performance improvement of a superconducting linear accelerator (SLA) system for a pellet injection to fuel the nuclear fusion reactor has been investigated numerically. For this purpose, a numerical code has been developed for analyzing the shielding current density and the dynamic motion of a high-temperature superconducting thin film by means of the FEM. In the present study, an exponentially increasing current is employed to accelerate a pellet in the container drastically. The results of the computations indicate that, compared to the results of the previous study, an acceleration distance in the SLA system to reach the target speed of 5 km/s is about 2.5 km and is decreased by a factor of about 0.12.

Near field communication (NFC) systems are based on the transmission of energy and information via an air interface between a polling and responding device. Both, the energy requirement and the information flow are defined in the underlying standard. This conformity to the standard must be proven accordingly for the respective applications in order to be allowed to bring NFC devices onto the market. The Q-factor of the overall system is an important system parameter for both power transmission and signal shape and consequently the information flow. Due to the fact that NFC systems have nonlinear electrical network components, the definition of this Q-factor is non-trivial and thus also complicates the definition of corresponding objective functions for the optimization of such NFC-systems. In this contribution, we therefore discuss the definition of the Q-factor in the case of nonlinear electrical circuits and its impact on the multimodal optimization of NFC systems.

Metamaterial (MM) is very promising in engineering applications since it exhibits extraordinary physical properties that do not exist in nature. To address the inefficiencies of existing multi-objective robust optimization methodologies in applications to MM designs, an improved multi-objective genetic algorithm and an adaptive response surface model are proposed. To accelerate the solution speed of the original multi-objective algorithm in finding both high quality solutions and distributing them uniformly, two polynomial approximation-based move operations are proposed. Moreover, some dominant techniques including the construction of the relationship between different objective functions, and the relationship between the objectives and the design variables are investigated. Also, an adaptive response surface model is introduced to efficiently quantify the robust performance of a solution. Optimization results of two mathematical benchmark problems and a prototype MM unit have demonstrated the feasibility and merits of the proposed methodology.

This paper proposes a design method for reducing torque ripple and high speed IPMSM used in motors for semiconductor CVD process. IPMSM can utilize the reluctance torque through the difference in inductance between the d-axis and the q-axis, but has a disadvantage in terms of torque ripple. Accordingly, an improved model was designed to reduce torque ripple and simultaneously reduce eddy current loss by applying taper ring to the rotor. The voltage and current limiting sources and speed-torque curves were compared with the existing model, and the no-load and load performance were compared by operating speed. In addition, the web thickness of the improved model was secured through demagnetization analysis and mechanical stiffness analysis of permanent mag-nets, and the design feasibility of the improved model and the existing model was verified through the FEA analysis results.

In this paper, a multi-objective topology optimization (MOTO) methodology using Non-dominated Sorting Genetic Algorithm II (NSGAII) and convolutional neural network (CNN) is proposed. The original NSGAII is improved to achieve better global search ability and uniform distribution of Pareto solutions. And CNN is applied as a surrogate model for finite element analysis. The framework of the proposed methodology is elaborated. To validate the proposed methodology, it is applied to the TO of an electromagnetic actuator. Numerical results demonstrate that computational cost of TO can be reduced without deteriorating the optimization quality.

This paper presents a topology optimization algorithm based on a Level-Set method and continuum sensitivity analysis in a 2D

magnetostatic finite element context. The novelty is the use of a fully analytically expression for the sensitivity analysis based on the

virtual work principle and applied to a Level-Set implementation. The numerical study of a switch reluctance machine and the virtual

work method to compute the torque are presented. Then, the Level-Set method is detailed in a general context. The corresponding

continuum sensitivity analysis using the adjoint method to maximize the torque is expressed. Finally, the application of this method

to the optimization of the rotor of a switch reluctance machine is shown.

The valve-side bushing of converter transformer has become one of the most important equipment. This paper studies the electric field distribution of the valve-side bushing under AC and DC voltage excitation, and analyzes the performance of the insulating medium inside the bushing, and the distortion effect of the space charge electric field formed in the bushing on the original electric field was analyzed. The results indicate that the distribution of the electric field inside the bushing is similar under AC and DC voltage excitation, but in the SF6 region filled in the capacitor core, there is a steeper voltage drop under DC voltage than AC excitation; However, space charges are generated in the DC electric field due to electric field polarization, and generate a space charge electric field that reacts against the DC voltage field, which has a great influence on the electric field distortion of the inner and outermost plates.

In this paper, a study was conducted on the rotor structure to increase the output and safety factor of the Synchronous Reluctance Motor (SynRM) for air conditioning in stations. First, to compensate for the low safety factor of SynRM, a shape that can increase the safety factor was adopted in the barrier structure. And in SynRM, design variables that can additionally increase the safety factor were adopted. A review of torque and safety factor according to these design variables was confirmed through finite element analysis. And through loss analysis, improvement design factors that can increase efficiency were found. Finally, the validity of the study was carried by comparing the analysis results with the experimental results of the prototype.