Online and Compositional Learning of Controllers with Application to Floor Heating

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

22 Citations (Scopus)

Abstract

Controller synthesis for stochastic hybrid switched systems, like e.g. a floor heating system in a house, is a complex computational task that cannot be solved by an exhaustive search though all the control options. The state-space to be explored is in general uncountable due to the presence of continuous variables (e.g. temperature readings in the different rooms) and even after digitization, the state-space remains huge and cannot be fully explored. We suggest a general and scalable methodology for controller synthesis for such systems. Instead of off-line synthesis of a controller for all possible input temperatures and an arbitrary time horizon, we propose an on-line synthesis methodology, where we periodically compute the controller only for the near future based on the current sensor readings. This computation is itself done by employing machine learning in order to avoid enumeration of the whole state-space. For additional scalability we propose and apply a compositional synthesis approach. Finally, we demonstrate the applicability of the methodology to a concrete floor heating system of a real family house.
Original languageEnglish
Title of host publicationTools and Algorithms for the Construction and Analysis of Systems : 22nd International Conference, TACAS 2016
Number of pages15
PublisherSpringer
Publication date2016
Pages244-259
ISBN (Print)978-3-662-49673-2
ISBN (Electronic)978-3-662-49674-9
DOIs
Publication statusPublished - 2016
Event22nd International Conference on Tools and Algorithms for the Construction and Analysis of Systems - Eindhoven, Netherlands
Duration: 2 Apr 20168 Apr 2016
Conference number: 22
http://www.etaps.org/2016/tacas

Conference

Conference22nd International Conference on Tools and Algorithms for the Construction and Analysis of Systems
Number22
CountryNetherlands
CityEindhoven
Period02/04/201608/04/2016
Internet address
SeriesLecture Notes in Computer Science
Volume9636
ISSN0302-9743

Fingerprint

Heating
Controllers
Analog to digital conversion
Concrete construction
Learning systems
Scalability
Temperature
Sensors

Cite this

Larsen, K. G., Mikučionis, M., Muniz, M., Srba, J., & Taankvist, J. H. (2016). Online and Compositional Learning of Controllers with Application to Floor Heating. In Tools and Algorithms for the Construction and Analysis of Systems: 22nd International Conference, TACAS 2016 (pp. 244-259). Springer. Lecture Notes in Computer Science, Vol.. 9636 https://doi.org/10.1007/978-3-662-49674-9_14
Larsen, Kim Guldstrand ; Mikučionis, Marius ; Muniz, Marco ; Srba, Jiri ; Taankvist, Jakob Haahr. / Online and Compositional Learning of Controllers with Application to Floor Heating. Tools and Algorithms for the Construction and Analysis of Systems: 22nd International Conference, TACAS 2016. Springer, 2016. pp. 244-259 (Lecture Notes in Computer Science, Vol. 9636).
@inproceedings{84a29a4567694312b9fa596f8f2009e8,
title = "Online and Compositional Learning of Controllers with Application to Floor Heating",
abstract = "Controller synthesis for stochastic hybrid switched systems, like e.g. a floor heating system in a house, is a complex computational task that cannot be solved by an exhaustive search though all the control options. The state-space to be explored is in general uncountable due to the presence of continuous variables (e.g. temperature readings in the different rooms) and even after digitization, the state-space remains huge and cannot be fully explored. We suggest a general and scalable methodology for controller synthesis for such systems. Instead of off-line synthesis of a controller for all possible input temperatures and an arbitrary time horizon, we propose an on-line synthesis methodology, where we periodically compute the controller only for the near future based on the current sensor readings. This computation is itself done by employing machine learning in order to avoid enumeration of the whole state-space. For additional scalability we propose and apply a compositional synthesis approach. Finally, we demonstrate the applicability of the methodology to a concrete floor heating system of a real family house.",
author = "Larsen, {Kim Guldstrand} and Marius Mikučionis and Marco Muniz and Jiri Srba and Taankvist, {Jakob Haahr}",
year = "2016",
doi = "10.1007/978-3-662-49674-9_14",
language = "English",
isbn = "978-3-662-49673-2",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "244--259",
booktitle = "Tools and Algorithms for the Construction and Analysis of Systems",
address = "Germany",

}

Larsen, KG, Mikučionis, M, Muniz, M, Srba, J & Taankvist, JH 2016, Online and Compositional Learning of Controllers with Application to Floor Heating. in Tools and Algorithms for the Construction and Analysis of Systems: 22nd International Conference, TACAS 2016. Springer, Lecture Notes in Computer Science, vol. 9636, pp. 244-259, 22nd International Conference on Tools and Algorithms for the Construction and Analysis of Systems , Eindhoven, Netherlands, 02/04/2016. https://doi.org/10.1007/978-3-662-49674-9_14

Online and Compositional Learning of Controllers with Application to Floor Heating. / Larsen, Kim Guldstrand; Mikučionis, Marius; Muniz, Marco; Srba, Jiri; Taankvist, Jakob Haahr.

Tools and Algorithms for the Construction and Analysis of Systems: 22nd International Conference, TACAS 2016. Springer, 2016. p. 244-259 (Lecture Notes in Computer Science, Vol. 9636).

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

TY - GEN

T1 - Online and Compositional Learning of Controllers with Application to Floor Heating

AU - Larsen, Kim Guldstrand

AU - Mikučionis, Marius

AU - Muniz, Marco

AU - Srba, Jiri

AU - Taankvist, Jakob Haahr

PY - 2016

Y1 - 2016

N2 - Controller synthesis for stochastic hybrid switched systems, like e.g. a floor heating system in a house, is a complex computational task that cannot be solved by an exhaustive search though all the control options. The state-space to be explored is in general uncountable due to the presence of continuous variables (e.g. temperature readings in the different rooms) and even after digitization, the state-space remains huge and cannot be fully explored. We suggest a general and scalable methodology for controller synthesis for such systems. Instead of off-line synthesis of a controller for all possible input temperatures and an arbitrary time horizon, we propose an on-line synthesis methodology, where we periodically compute the controller only for the near future based on the current sensor readings. This computation is itself done by employing machine learning in order to avoid enumeration of the whole state-space. For additional scalability we propose and apply a compositional synthesis approach. Finally, we demonstrate the applicability of the methodology to a concrete floor heating system of a real family house.

AB - Controller synthesis for stochastic hybrid switched systems, like e.g. a floor heating system in a house, is a complex computational task that cannot be solved by an exhaustive search though all the control options. The state-space to be explored is in general uncountable due to the presence of continuous variables (e.g. temperature readings in the different rooms) and even after digitization, the state-space remains huge and cannot be fully explored. We suggest a general and scalable methodology for controller synthesis for such systems. Instead of off-line synthesis of a controller for all possible input temperatures and an arbitrary time horizon, we propose an on-line synthesis methodology, where we periodically compute the controller only for the near future based on the current sensor readings. This computation is itself done by employing machine learning in order to avoid enumeration of the whole state-space. For additional scalability we propose and apply a compositional synthesis approach. Finally, we demonstrate the applicability of the methodology to a concrete floor heating system of a real family house.

U2 - 10.1007/978-3-662-49674-9_14

DO - 10.1007/978-3-662-49674-9_14

M3 - Article in proceeding

SN - 978-3-662-49673-2

T3 - Lecture Notes in Computer Science

SP - 244

EP - 259

BT - Tools and Algorithms for the Construction and Analysis of Systems

PB - Springer

ER -

Larsen KG, Mikučionis M, Muniz M, Srba J, Taankvist JH. Online and Compositional Learning of Controllers with Application to Floor Heating. In Tools and Algorithms for the Construction and Analysis of Systems: 22nd International Conference, TACAS 2016. Springer. 2016. p. 244-259. (Lecture Notes in Computer Science, Vol. 9636). https://doi.org/10.1007/978-3-662-49674-9_14