Physics-Guided Neural Networks for Feedforward Control

Physics-Guided Neural Networks for Feedforward Control: An Orthogonal Projection-Based Approach

Kon, Johan (Eindhoven University of Technology)
Bruijnen, Dennis (Philips Research)
van de Wijdeven, Jeroen (ASML)
Heertjes, Marcel (Eindhoven University of Technology; ASML)
Oomen, T.A.E. (TU Delft Team Jan-Willem van Wingerden; Eindhoven University of Technology)

2022

Unknown nonlinear dynamics can limit the performance of model-based feedforward control. The aim of this paper is to develop a feedforward control framework for systems with unknown, typically nonlinear, dynamics. To address the unknown dynamics, a physics-based feedforward model is complemented by a neural network. The neural network output in the subspace of the model is penalized through orthogonal projection. This results in uniquely identifiable model coefficients, enabling increased performance and similar task flexibility with respect to the model-based controller. The feedforward framework is validated on a representative system with performance limiting nonlinear friction characteristics.

Training
Maximum likelihood detection
Limiting
System dynamics
Nonlinear filters
Cost function
Feedforward neural networks

http://resolver.tudelft.nl/uuid:b0513881-b2e7-4985-bd22-0f49a121e6b8

https://doi.org/10.23919/ACC53348.2022.9867653

IEEE

2023-03-05

978-1-6654-5196-3

Proceedings of the American Control Conference (ACC 2022)

2022 American Control Conference, ACC 2022, 2022-06-08 → 2022-06-10, Atlanta, United States

Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Institutional Repository

conference paper

© 2022 Johan Kon, Dennis Bruijnen, Jeroen van de Wijdeven, Marcel Heertjes, T.A.E. Oomen