Please use this identifier to cite or link to this item: http://hdl.handle.net/11452/22837
Title: Prediction of power losses in transformer cores using feed forward neural network and genetic algorithm
Authors: Uludağ Üniversitesi/Fen-Edebiyat Fakültesi/Fizik Bölümü.
0000-0003-2546-0022
Küçük, İlker
Derebaşı, Naim
AAI-2254-2021
6602910810
11540936300
Keywords: Engineering
Instruments & instrumentation
Toroidal magnetic cores
Power loss
Genetic algorithm
Artificial neural network
Optimization
Neural networks
Mathematical models
Magnetization
Genetic algorithms
Computational geometry
Toroidal magnetic cores
Power loss
Geometrical effects
Electric transformers
Frequency
Magnetic-properties
Toroidal cores
Issue Date: Aug-2006
Publisher: Elsevier
Citation: Küçük, İ. ve Derebaşı, N. (2006). ''Prediction of power losses in transformer cores using feed forward neural network and genetic algorithm''. Measurement: Journal of the International Measurement Confederation, 39(7), 605-611.
Abstract: A mathematical model for core losses was improved for frequency and geometrical effects using experimental data obtained from toroidal wound cores. The improved mathematical model was applied to the other soft magnetic materials and optimizes its parameters with the aim of neural networks. A 6-neuron input layer, 9-neuron output layer model with two hidden layers were developed. While the input neurons were geometrical parameters, magnetising frequency, magnetic induction and resistivity of the soft magnetic materials, output neurons were correlation coefficients and the power loss. The network has been trained by the genetic algorithm. The linear correlation coefficient was found to be 99%.
URI: https://doi.org/10.1016/j.measurement.2006.02.001
https://www.sciencedirect.com/science/article/pii/S0263224106000212
http://hdl.handle.net/11452/22837
ISSN: 0263-2241
Appears in Collections:Web of Science

Files in This Item:
There are no files associated with this item.


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.