Multi-State Modelling and Observation of Magneto-Rheological Clutch With Rate-Dependent Hysteresis Characteristic
Guangzeng Chen, Xiaogang Xiong, Yunjiang Lou
Abstract
Magneto-rheological clutches (MRCs) are potential for providing jointed robots with the capability of behaving compliantly. However, due to the nonlinear dynamics of rate-dependent hysteresis exhibited by the MRC, precisely controlling its behaviors is still cumbersome. This paper firstly proposes a multi-state MRC model described by multiple ordinary differential equations (ODEs) that captures the complicated rate-dependent hysteresis phenomenon. The model offers better estimation accuracy of rate-dependent hysteresis compared to the Bouc-Wen model. Then, to further improve the modelling performance in various inputs, this paper designs a robust and exact observer based on the super-twisting algorithm (STA) to compensate the model uncertainties. It exactly estimates the output torque of MRC by compensating for the model uncertainties, even when the parameters of the MRC model are not optimized. Comparisons and experiments demonstrate the robustness and estimation accuracy of the STA observer.
BibTeX
@inproceedings{ral2021_multistatemodell,
title = {Multi-State Modelling and Observation of Magneto-Rheological Clutch With Rate-Dependent Hysteresis Characteristic},
author = {Guangzeng Chen and Xiaogang Xiong and Yunjiang Lou},
booktitle = {RA-L 2021},
year = {2021}
}