Optimal design of hollow conductor for high‐speed synchronous motor exploiting adaptive‐sampling radial basis function algorithm
Abstract As aircraft electrification advances, permanent magnet synchronous motors (PMSMs) require higher power density and efficiency, but optimisation is hindered by high computational costs and resource consumption. To address this, the paper proposes a multi‐objective optimisation method based o...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2024-12-01
|
| Series: | IET Electric Power Applications |
| Subjects: | |
| Online Access: | https://doi.org/10.1049/elp2.12509 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract As aircraft electrification advances, permanent magnet synchronous motors (PMSMs) require higher power density and efficiency, but optimisation is hindered by high computational costs and resource consumption. To address this, the paper proposes a multi‐objective optimisation method based on adaptive sampling radial basis function (ASRBF). The ASRBF algorithm adaptively adds sample points by estimating expected improvements at prediction points, enabling the surrogate model to rapidly approximate the global optimum while significantly reducing function evaluations. It integrates optimisation objectives and constraints using probabilistic improvement techniques, enhancing robustness and convergence speed by avoiding excessive exploration of invalid regions. Mathematical test functions validate ASRBF's excellent performance in handling complex objective domains. Applied to high‐speed PMSM with hollow conductors, it aims to minimise AC losses while maximising slot fill factor and heat dissipation, resulting in a 15% reduction in losses and an increase in conductor heat dissipation area and slot fill factor, at one‐thousandth of the cost of the full factorial optimisation method. The ASRBF algorithm efficiently constructs surrogate models for multi‐dimensional, multi‐objective, non‐linear, and constrained problems, providing a powerful tool for comprehensive performance optimisation of complex systems such as motors. |
|---|---|
| ISSN: | 1751-8660 1751-8679 |