Probabilistic Physics-Augmented Neural Networks for Robust Control Applied to a Slider-Crank System

Edward Kikken, Jeroen Willems, Rob Salaets, Erik Hostens

2023

Abstract

Key industrial trends such as increasing energy and performance requirements as well as mass customization, lead to more complex non-linear machines with many variants. For such systems, variability in dynamics arising from many factors significantly affects performance. To capture this adequately we introduce a probabilistic extension of a Physics-Augmented Neural Network (PANN). We subsequently illustrate the added value of such models in robust optimal control, thereby keeping performance high while guaranteeing to meet the application’s constraints. The approach is validated on a experimental slider-crank mechanism, which is ubiquitous in industrial machines.

Paper Citation

in Harvard Style

Kikken E., Willems J., Salaets R. and Hostens E. (2023). Probabilistic Physics-Augmented Neural Networks for Robust Control Applied to a Slider-Crank System. In Proceedings of the 20th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO; ISBN 978-989-758-670-5, SciTePress, pages 152-161. DOI: 10.5220/0012168200003543

in Bibtex Style

@conference{icinco23,
author={Edward Kikken and Jeroen Willems and Rob Salaets and Erik Hostens},
title={Probabilistic Physics-Augmented Neural Networks for Robust Control Applied to a Slider-Crank System},
booktitle={Proceedings of the 20th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO},
year={2023},
pages={152-161},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012168200003543},
isbn={978-989-758-670-5},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 20th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO
TI - Probabilistic Physics-Augmented Neural Networks for Robust Control Applied to a Slider-Crank System
SN - 978-989-758-670-5
AU - Kikken E.
AU - Willems J.
AU - Salaets R.
AU - Hostens E.
PY - 2023
SP - 152
EP - 161
DO - 10.5220/0012168200003543
PB - SciTePress