A numerical approach for the evaluation of hysteresis loops in the harmonic regime is presented. Genetic algorithms (GAs) are used to train neural networks (NNs) with the aim of generalizing the Jiles-Atherton (JA) static hysteresis model for dynamic loops. The NN training is based on symmetrical and asymmetrical, major and minor loops under sinusoidal excitation with or without offset. Subsequently, the harmonic magnetic time period has been partitioned into suitable time windows into which the field has been fitted by sinusoids with offset. New JA parameters, estimated by the trained NNs in each partitioning time window, have been inserted into the JA static model to calculate the magnetization waveform, time window by time window. Validations are shown.
RIGANTI FULGINEI, F., Salvini, A., Coltelli, C. (2003). A Neuro-Genetic and Time-Frequency Approach Macromodeling Dynamic Hysteresis in Harmonic Regime. IEEE TRANSACTIONS ON MAGNETICS, 39, 1401-1404 [10.1109/TMAG.2003.810539].
A Neuro-Genetic and Time-Frequency Approach Macromodeling Dynamic Hysteresis in Harmonic Regime
RIGANTI FULGINEI, Francesco;SALVINI, Alessandro;
2003-01-01
Abstract
A numerical approach for the evaluation of hysteresis loops in the harmonic regime is presented. Genetic algorithms (GAs) are used to train neural networks (NNs) with the aim of generalizing the Jiles-Atherton (JA) static hysteresis model for dynamic loops. The NN training is based on symmetrical and asymmetrical, major and minor loops under sinusoidal excitation with or without offset. Subsequently, the harmonic magnetic time period has been partitioned into suitable time windows into which the field has been fitted by sinusoids with offset. New JA parameters, estimated by the trained NNs in each partitioning time window, have been inserted into the JA static model to calculate the magnetization waveform, time window by time window. Validations are shown.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
