Noise Handling in Data-driven Predictive Control: A Strategy Based on Dynamic Mode Decomposition

Andrea Sassella, Valentina Breschi, Simone Formentin
Proceedings of The 4th Annual Learning for Dynamics and Control Conference, PMLR 168:74-85, 2022.

Abstract

A major issue when exploiting data for direct control design is noise handling, since overlooking or improperly treating noise might have a catastrophic impact on closed-loop performance. Nonetheless, standard approaches to mitigate its effect might not be easily applicable for data-driven control design, since they often require tuning a set of hyper-parameters via potentially unsafe closed-loop experiments. By focusing on data-driven predictive control, we propose a noise handling approach based on truncated dynamic mode decomposition, along with an automatic tuning strategy for its hyper-parameters. By leveraging on pre-processing only, the proposed approach allows one to avoid dangerous closed-loop calibrations while being effective in coping with noise, as illustrated on a benchmark simulation example.

Cite this Paper

BibTeX
@InProceedings{pmlr-v168-sassella22a, title = {Noise Handling in Data-driven Predictive Control: A Strategy Based on Dynamic Mode Decomposition}, author = {Sassella, Andrea and Breschi, Valentina and Formentin, Simone}, booktitle = {Proceedings of The 4th Annual Learning for Dynamics and Control Conference}, pages = {74--85}, year = {2022}, editor = {Firoozi, Roya and Mehr, Negar and Yel, Esen and Antonova, Rika and Bohg, Jeannette and Schwager, Mac and Kochenderfer, Mykel}, volume = {168}, series = {Proceedings of Machine Learning Research}, month = {23--24 Jun}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v168/sassella22a/sassella22a.pdf}, url = {https://proceedings.mlr.press/v168/sassella22a.html}, abstract = {A major issue when exploiting data for direct control design is noise handling, since overlooking or improperly treating noise might have a catastrophic impact on closed-loop performance. Nonetheless, standard approaches to mitigate its effect might not be easily applicable for data-driven control design, since they often require tuning a set of hyper-parameters via potentially unsafe closed-loop experiments. By focusing on data-driven predictive control, we propose a noise handling approach based on truncated dynamic mode decomposition, along with an automatic tuning strategy for its hyper-parameters. By leveraging on pre-processing only, the proposed approach allows one to avoid dangerous closed-loop calibrations while being effective in coping with noise, as illustrated on a benchmark simulation example.} }
Endnote
%0 Conference Paper %T Noise Handling in Data-driven Predictive Control: A Strategy Based on Dynamic Mode Decomposition %A Andrea Sassella %A Valentina Breschi %A Simone Formentin %B Proceedings of The 4th Annual Learning for Dynamics and Control Conference %C Proceedings of Machine Learning Research %D 2022 %E Roya Firoozi %E Negar Mehr %E Esen Yel %E Rika Antonova %E Jeannette Bohg %E Mac Schwager %E Mykel Kochenderfer %F pmlr-v168-sassella22a %I PMLR %P 74--85 %U https://proceedings.mlr.press/v168/sassella22a.html %V 168 %X A major issue when exploiting data for direct control design is noise handling, since overlooking or improperly treating noise might have a catastrophic impact on closed-loop performance. Nonetheless, standard approaches to mitigate its effect might not be easily applicable for data-driven control design, since they often require tuning a set of hyper-parameters via potentially unsafe closed-loop experiments. By focusing on data-driven predictive control, we propose a noise handling approach based on truncated dynamic mode decomposition, along with an automatic tuning strategy for its hyper-parameters. By leveraging on pre-processing only, the proposed approach allows one to avoid dangerous closed-loop calibrations while being effective in coping with noise, as illustrated on a benchmark simulation example.
APA
Sassella, A., Breschi, V. & Formentin, S.. (2022). Noise Handling in Data-driven Predictive Control: A Strategy Based on Dynamic Mode Decomposition. Proceedings of The 4th Annual Learning for Dynamics and Control Conference, in Proceedings of Machine Learning Research 168:74-85 Available from https://proceedings.mlr.press/v168/sassella22a.html.

Related Material