Experimental Study and Modular Modeling of Magnetostrictive Hysteresis With Data-Driven Approach

Abstract

A new modular modeling approach to describe the magnetostrictive hysteresis is presented in this paper. The hysteresis exhibits varying dynamics under the input signals of the different frequencies and amplitudes. To this end, the experimental characterization is conducted and hysteresis modeling approach is studied with the data-driven technique. Two characteristic indexes, i.e., loop relative width and loop asymmetry coefficients, are quantitatively analyzed. Based on the hysteresis phenomenon analyses, different modular models are selected and combined to describe those phenomena. The arctangent-polynomial modified Prandtl-Ishlinskii (APMPI) and infinite impulse response (IIR) submodels are applied for asymmetry rate-independent and rate-dependent hysteresis identification respectively. Those submodels are selected to construct a cascaded overall model to describe the hysteresis of magnetostrictive actuator. The experimental results demonstrate that, with the proposed hysteresis modular modeling approach, better performance can be obtained than some other approaches in terms of modeling accuracy.