Alireza Roosta - Shiraz University of Technology, Electrical and Electronics Engineering Department
Behrouz Safarinejadian - Shiraz University of Technology, Electrical and Electronics Engineering Department
Alireza Soltanimehr - Shiraz University of Technology, Electrical and Electronics Engineering Department

One of the main issues in modeling the behaviorof electric machineries is the process of modeling themagnetic material used in such machineries. Almostall ferromagnetic materials display a set of behaviorswhich are known under the term of magnetichysteresis. Models which are proposed for thehysteresis phenomenon based on the physicalbehavior of ferromagnetic materials, such asPreisach, is so complicated which require largecomputer storage and also consume a lot of time forcalculation. Therefore, artificial neural networkscould be considered as suitable alternatives sincealong with featuring a high accuracy, they are fastand require less computer storage. Using multi-layerfeed-forward neural networks in the following paper,a model for magnetic hysteresis is proposed whichcould model all internal loops along with the mainhysteresis loop. Results to the simulation suggest agood agreement between the aforementioned modeland accurate model of Preisach

Hysteresis identification, Preisach model, neural network, perceptron, radial basis function (RBF)