The present investigation aims at the definition of an efficient and robust neural network-based model to simulate the magnetic hysteresis in performing magnetic alloys suitable for aircraft applications. Starting from a set of measured hysteresis loops, a convenient and effective method to train the network consists to identify the Preisach model and use it for the generation of the training set. The obtained neural network turned out to be particularly robust and able to reproduce the behaviour of the Preisach model with a significant reduction of the computational time. The comparative analysis between the two approaches takes into account different kinds of excitation waveforms.
Quondam Antonio, S., Riganti Fulginei, F., Rimal, H.P., Ghanim, A.M. (2020). On the Use of Feedforward Neural Networks to Simulate Magnetic Hysteresis in Electrical Steels. In 20th IEEE Mediterranean Electrotechnical Conference, MELECON 2020 - Proceedings (pp.119-124). Institute of Electrical and Electronics Engineers Inc. [10.1109/MELECON48756.2020.9140585].
On the Use of Feedforward Neural Networks to Simulate Magnetic Hysteresis in Electrical Steels
Riganti Fulginei F.;
2020-01-01
Abstract
The present investigation aims at the definition of an efficient and robust neural network-based model to simulate the magnetic hysteresis in performing magnetic alloys suitable for aircraft applications. Starting from a set of measured hysteresis loops, a convenient and effective method to train the network consists to identify the Preisach model and use it for the generation of the training set. The obtained neural network turned out to be particularly robust and able to reproduce the behaviour of the Preisach model with a significant reduction of the computational time. The comparative analysis between the two approaches takes into account different kinds of excitation waveforms.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
