Improved EHM-based NN hysteresis model

An improved EHM-based hysteresis model is proposed in this paper. In this scheme, neural networks are employed to implement the mapping between the input and output of hysteresis. The so-called elementary hysteresis model (EHM) is introduced to construct the transformation to transform the multi-valued mapping of hysteresis into a one-to-one mapping. In order to construct the EHM, parabolas are chosen as monotone curves in the EHM-based NN model. In this paper, a new method for determining the coefficients of the parabolas is proposed. The coefficients of the parabolas are different from each other in the improved EHM-based NN model. Finally, both the numerical and experimental examples of using the proposed modeling scheme are presented.