An Improved Prediction Algorithm for Noise of Transformer Considering Material Parameter Uncertainty

This study proposes a rapid transformer noise prediction method to quantify the propagation of electrical steel material uncertainties in transformer systems. To characterize uncertainties in silicon steel, stochastic models of magnetization and magnetostriction behaviors are developed using vine co...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE transactions on magnetics s. 1
Hlavní autoři: Yang, Fan, Xia, Yisha, Wang, Jiawei, Wang, Pengbo, Jiang, Hui
Médium: Journal Article
Jazyk:angličtina
Vydáno: IEEE 2025
Témata:
Adaptation models
backpropagation algorithms
Computational modeling
Magnetic material
Magnetization
Magnetostriction
Noise
noise level
Power transformer insulation
Stochastic processes
stochastic theory
Transformers
Uncertainty
Vectors
ISSN:0018-9464, 1941-0069
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:This study proposes a rapid transformer noise prediction method to quantify the propagation of electrical steel material uncertainties in transformer systems. To characterize uncertainties in silicon steel, stochastic models of magnetization and magnetostriction behaviors are developed using vine copula theory, generating correlated sample curves across flux densities ranging from 0.1-1.9 T. To mitigate computational costs, an adaptive differential evolution-optimized backpropagation neural network (ADE-BP) is introduced for transformer noise analysis. The proposed method is implemented to the finite element model of a transformer to validate its effectiveness and accuracy. Numerical results reveal a 12.7% probability of exceeding permissible noise thresholds due to variations in magnetic parameters, thereby underscoring the necessity for topology optimization in high-efficiency, low-noise transformer design methodologies.
ISSN:0018-9464
1941-0069
DOI:10.1109/TMAG.2025.3626932