Conference Agenda

Although the output power of an induction motor (IM) is generally inferior to that of an interior permanent magnet synchronous motor, IMs are widely applied to industrial machines owing to their solidity, lower cost, self-starting, and so on. When high-performance IM design is carried out, a virtual approach using the finite-element method is the usual practice. However, because a finite-element analysis of an IM takes a long time owing to the long transient state in the time domain, it is difficult to perform shape optimization and topology optimization based on evolutionary algorithms. To overcome this difficulty, the sensitivity-based topology optimization method of IMs in the time domain was proposed.

Since the interior permanent magnet synchronous motor (IPMSM) is essential for the driving of the hybrid-car and electric vehicle, the design of highly efficient IPMSM is enthusiastically requested by the industrial world. When the effective design of IPMSM is schemed, the topology optimization procedure is frequently applied to the determination of the magnetic structure of IPMSM. The accurate evaluation of IPMSM performance in the time domain is crucial for optimization enhancement. However, the target of topology optimization based on the sensitivity analysis using the adjoint variable method is focused on the magnetostatic filed. In this paper, to extend the topology optimization of IPMSM in magnetostatic field to the time domain, the sensitivity analysis using adjoint variable method supported by finite element analysis coupled with three-phase voltage source is proposed.

This paper proposes an efficient surrogate-assisted design optimization method for a surface-mounted permanent-magnet vernier machine (SPMVM) based on subdomain analysis. Combined with the Saltelli sampling method, magnetic fields, torque characteristics, and power factor of the SPMVM are analytically predicting by subdomain models. A surrogate model with high generalization ability is trained by the extreme learning machine (ELM) with the subdomain analysis results. An optimal design that yields higher average torque, low torque ripple, and higher power factor is searched by non-dominated sorting genetic algorithm II (NSGAII).

The design specification of future high-field superconducting magnets is pushing the low temperature superconductors towards their ultimate performance. Numerical models support each design stage from selection of materials, through the design of cable geometry to magnet geometry definition. In order to meet stringent design objectives, a multi-model and multi-scale optimization is performed. In this paper, we present a model-based workflow for superconducting magnet design optimization. To this aim, we incorporate relevant model-based system engineering concepts in order to streamline the design process. Furthermore, the advances in software engineering allow for encoding the design process as a workflow composed of interactive notebooks incorporating each design step. Once a design workflow is established, its execution is automated to ensure design reproducibility and traceability. We demonstrate the proposed methodology with a geometry design optimization of a high-field superconducting magnet.

In this paper, a forward uncertainty propagation method is proposed for a 2D finite element method (FEM) of an induction machine. This method is used to quantify the uncertainty of input parameters (e.g., dimensions, material properties), to demonstrate their variability effect on broken rotor bar fault harmonics. In order to show the most influential input parameters to the broken rotor bar fault harmonics, a global sensitivity analysis from the polynomial chaos expansion (PCE) approximation of the FEM model will be studied. This method can be applicable for minimizing the amplitude deviation of the BRB fault harmonics between the finite element model and measurement data.