Optimum Identification of Iron Loss Models in NGO Electrical Steel for Power Electronics

S. Quondam Antonio
DOI: https://doi.org/10.1109/rtsi.2019.8895589
2019-09-01
Abstract:Prediction and modelling of the dynamic power losses, and their decomposition into the hysteresis and eddy current contributions, in laminated ferromagnetic materials is not a comprehensively solved problem. This paper mostly deals with the experimental identification of Preisach-type models, aiming at an accurate and effective reproduction of the hysteresis losses. Different probability density functions are examined to approximate the distribution of the hysteresis operators. The distribution parameters are identified exploiting both a genetic algorithm, suitably developed, and the patternsearch approach. The comparison of the results obtained for a NGO electrical steel, particularly appropriate in power electronics applications, seemed to indicate a single Lorentzian function as the best-fit hysteron distribution. The hysteresis losses evaluated by the Preisach model turned out to be effective also when the material is excited by non sinusoidal polarization waveforms.
What problem does this paper attempt to address?