The paper presents a measurement system which can be used to prepare measured data to identify the Preisach hysteresis model. The identification is based on the Everett function. A fast and memory efficient realization of the model will be shown which is very useful in numerical field analysis techniques. The dynamic extension of the static model is based on the loss separation formula, and the measured excess magnetic field has been approximated by a feed-forward type neural network. The developed model has been inserted into a finite element procedure.