Predictive modeling and multi-objective optimization of magnetic core loss with activation function flexibly selected Kolmogorov-Arnold networks

Yanzhan Chen, Fan Yu, Li Chen, Ge Jin and Qian Zhang

Energy, 2025, vol. 334, issue C

Abstract: Magnetic core loss is inevitably present in electromagnetic devices such as power equipment and electric motors, and it directly impacts their energy efficiency. Therefore, reducing magnetic core loss is an urgent task that can significantly decrease energy consumption on a global scale. However, different magnetic core materials exhibit distinct loss characteristics, and each material potentially demonstrates significantly different performance under specific conditions. To address this variability, this paper proposes an activation function flexibly selected Kolmogorov-Arnold network (FS-KAN) to develop a universal magnetic core loss model. This model comprehensively encompasses various materials and operating conditions while ensuring high predictive accuracy. Our proposed FS-KAN employs an adaptive strategy to choose the optimal activation mode at each feedforward KAN node and outperforms baseline methods in the magnetic core loss modeling tasks. Furthermore, an interpretable machine learning framework for core loss modeling is developed, revealing the complex interplay of multiple factors such as the magnetic properties of the materials, operating conditions (such as frequency, temperature, and waveform), and magnetic flux density. Finally, employing FS-KAN as a surrogate model, we establish a bi-objective mixed-integer optimization model and employ several popular heuristic algorithms to determine the optimal operating conditions for magnetic components. The obtained Pareto front exhibits excellent uniformity and completeness, validating the effectiveness of the proposed “learn and optimize” approach for magnetic core loss analysis.

Keywords: Core loss modeling; Kolmogorov-Arnold networks; Adaptive activation function; Interpretable machine learning; Multi-objective optimization (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544225033729
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:334:y:2025:i:c:s0360544225033729

DOI: 10.1016/j.energy.2025.137730

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().