Conference Agenda

The design of inductors for power electronics converters aims to obtain a design differential inductance value together with low losses and high volume power-density values. However, these are often contrasting objectives. The related multi-objective optimisation problem can be effectively tackled through population-based algorithms, such as Artificial Immune Systems. As these approaches require to evaluate many designs through time-consuming procedures, a classifier system trained in advance to recognise non-admissible solutions can support the search of candidate solutions. Different strategies in realising the classifier are presented and discussed.

This paper presents a topology optimization of a surface permanent magnet (SPM) motor. The present topology optimization method determines not only the material distribution but also magnetization direction in PMs to maximize the torque density at a constant current. It is shown that the torque performance of the optimized motor is superior to that of the conventional motor with Halbach array whose magnetization linearly varies along the circumferential direction. The optimized motor is novel in that the triangular magnetic material is placed on the rotor surface.

The optimization of electrical machine topologies is an area of continuous research, spanning a range of topologies, technologies, and techniques. One such example is that of the synchronous reluctance machine, with the rotor often the subject of topology optimization. This paper builds upon prior work by developing a directed optimization approach that explores geometric formations to improve computational efficiency and more thoroughly explore the design space. These methods are further executed on a tailored cloud computing platform to leverage large scale computing power, without requiring modification to the underlying optimization code. This enables very large, multi-dimensional problems to be solved quickly. Results demonstrate that the design space could be explored more fully and efficiently using the directed optimization approach when compared to conventional optimization approaches. Further work is identified as refinement of the optimization parameters and investigating use of the generated dataset to train machine learning algorithms through structured domain randomization.

In this paper, we investigate the synthesis of antenna arrays capable of producing different radiation patterns, which can also be easily manufactured using planar Printed Circuit Board (PCB) technology. To this end, we employ algorithms using simultaneous

placement techniques of antennas into non-regular arrays. We also inspect the effect of symmetries in the specification upon the symmetries of the array’s geometry. The arrays are simulated via the Finite Element Method (FEM). Through numerical examples we illustrate the scope and convergence of the method. Applications of interest include circularly polarized patterns, since the algorithm can handle the rotation as well as translation of the antenna elements, in addition to broadband systems, as the method is capable of characterizing the performance of arrays at multiple frequencies.

In this paper, a robust topology optimization methodology for designing the shape of a permanent magnet synchronous motor that can satisfy the target torque performance in the occurrence of unpredictable rotor eccentricity is presented. The static eccentricity of the rotor, which is a mechanical uncertainty that may occur during the motor manufacturing process, is considered as a major factor affecting torque performances, and a robust motor shape that can achieve target output power and reduce torque ripple under eccentricity is optimized. By predicting the distortion of the airgap magnetic flux distribution by the degree of eccentricity in advance and setting the target magnetic flux distribution for improving torque performances, it is possible to optimize the motor shape considering the eccentricity even when using an efficient periodic model rather than a full-scale model. The outer boundaries of the rotor, and the shape of the permanent magnet are represented by the level set functions, and the area near airgap is set as the design domain to perform the motor design. Several optimizations are performed by setting the probability and degree of eccentricity in various ways, and representative features of the optimized motor shape, derived when considering eccentricity, are presented.

This paper presents a topology optimization of a microstrip bandpass filter to obtain novel circuit structure. The conductor shape is represented by the weighted sum of the normalized Gaussian functions whose weighting coefficients are determined by CMA-ES. It is shown that the scattering parameters obtained by the optimization agree well with the measured results. Moreover, the proposed method provides a wide pass band which is composed of multiple poles.

This paper presents a novel multi objective optimization method for hybrid-field motors. In the proposed method, the optimization process is divided into two steps. In the first step, combinatorial optimization is performed to obtain solutions with a geometry that combines the structure of the wound field motor and interior permanent magnet motor. In the second step, a local search is performed within the constrained design region for each Pareto solution obtained in the first step. To validate the effectiveness of the proposed method, it is applied to the optimization problem of a hybrid-field motor.

Nonlinear field grading materials protect high voltage direct current cable joints from dielectric damage. Their development is a challenging task, due to a large number of material parameters. This work introduces the adjoint variable method as an efficient approach to study the sensitivities of material parameters. The method is is applied to a 320 kV-cable joint, subjected to a lightning impulse. The results are compared to the direct sensitivity method. Both time-integrated quantities of interest as well as quantities evaluated at a specific point in time are considered.

Concentrated IPMSM has advantages in various aspects, such as space ratio, mass productivity, and copper loss, but concentrated winding has a limitation that the eddy current loss of permanent magnets is quite large compared to distributed winding. Eddy current loss means the heat of the rotor and becomes larger during high-speed operation. Therefore, in the concentrated IPMSM that requires high-speed operation, it is essential to design to reduce eddy current loss. In this paper, in order to reduce the eddy current loss of the IPMSM, a shape that increases the reluctance of the q-axis was applied to the rotor, and each was designed by dividing into concentrated winding and distributed winding. The number of pole slots of concentrated winding was selected as 6 poles and 9 slots, and the number of pole slots of distributed winding was selected as 6 poles and 27 slots. Through FEA, the voltammetry limit circle and T-N curve were compared according to each winding method. By comparing the d-axis inductance and q-axis inductance, a model that is advantageous at high speed was confirmed, and the no-load and load characteristics according to the operating speed were compared in detail.

This paper shows the useful combination of a gradient-based particle swarm optimization (GPSO) method with a metamodeling process in order to save computation time for the design of a Wireless Power Transfer (WPT) system for automotive applications. The goal of this analysis has been to investigate new configurations for 3F3 ferrite cores in an existing WPT system regarding both the coupling factor and the ferrite volumes. An innovative gradient-based multi-objective optimization method has been coupled to an adaptive sampling algorithm for Polynomial-Chaos Kriging metamodeling.

In this paper, we analyzed electromagnetic performance and axial electromagnetic force according to design variables of axial flux permanent magnet synchronous motor (AFPMM). Unlike radial flux permanent magnet synchronous motor (RFPMM), AFPMM generates axial electromagnetic force. It deteriorates motor performance, vibration, and noise characteristic. So we need to design for reduction in the axial electromagnetic force. Therefore, we conduct sensitivity analysis of electromagnetic performance and axial electromagnetic force according to design variables. After the sensitivity analysis, improvement design for axial electromagnetic force is performed by applying sensitivity analysis results. The proposed axial electromagnetic force improvement design is analyzed using a three-dimensional finite element analysis (3D-FEA) and virtual work principle (VWP).