摘要

In this paper, the concept of constraint factor (CF) is proposed and introduced into the construction of hysteretic operator (HO). The parameters of HO are computed by the normalization and least square methods. And then, based on the constructed HO, the input space is expanded from 1-dimension to 2-dimension. The mapping between the new input and output spaces is comprised of one-to-one and multiple-to-one mappings which can be identified using neural networks. Finally, in the experiment, a neural network is employed to model hysteresis for the magnetostrictive actuator. The experimental results approve the proposed approach.