Experimental data for An Hysteresis Model called MFGR Neural Network

Published: 17 October 2023| Version 1 | DOI: 10.17632/k924fhc2xs.1
Geng Wang


test data for An Accurate and Interpretable Rate-Dependent Asymmetric Hysteresis Model by integrating Magic Formula with Generalized Regression Neural Network


Steps to reproduce

The experimental setup consists of a computer equipped with MATLAB software, a dSPACE featuring 16-bit A/D and D/A converters, a piezoelectric tip/tilt platform, a power amplifier, and a built-in strain displacement sensor. The data acquisition process follows these steps: the dSPACE controller generates a voltage signal with a specific amplitude as per the command from the upper computer; the power amplifier amplifies this voltage signal and transmits it to the piezoelectric platform, inducing a corresponding deflection motion; the built-in displacement sensor measures the angular displacement, which is then collected by the A/D converter and stored in the upper computer for further processing. The applied voltage signal and the resulting measured displacement signal serve as the basis for modeling the nonlinear hysteresis response of the piezoelectric platform.


Southwest University of Science and Technology


Mechanical Engineering