Model-free finite-time H2/H∞ predictive control for discrete-time systems via Q-learning

Yihong Lin, Haiying Wan, Peng He, Xiaoli Luan and Fei Liu

International Journal of Systems Science, 2025, vol. 56, issue 15, 3718-3729

Abstract: This paper proposes a model-free finite-time $ H_2/H_\infty $ H2/H∞ predictive control strategy for linear discrete-time systems with unknown parameters. The method employs Q-learning to learn the optimal control policy directly from measured input-output data, eliminating the need for an explicit system model. The developed control strategy aims to achieve a finite-time $ H_2/H_\infty $ H2/H∞ performance index, balancing robustness and the energy consumption of the control input in a given intervals, thereby enhancing the overall system performance. The algorithm incorporates receding horizon optimisation to dynamically adjust the control strategy the evolving disturbances. Therefore, the robustness of the considered system is further improved. Simulation results demonstrate the effectiveness of the proposed method, showcasing its potential for practical applications in various control systems, particularly those with limited model information or model uncertainties.

Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://hdl.handle.net/10.1080/00207721.2025.2475360 (text/html)
Access to full text is restricted to subscribers.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:taf:tsysxx:v:56:y:2025:i:15:p:3718-3729

Ordering information: This journal article can be ordered from
http://www.tandfonline.com/pricing/journal/TSYS20

DOI: 10.1080/00207721.2025.2475360

Access Statistics for this article

International Journal of Systems Science is currently edited by Visakan Kadirkamanathan

More articles in International Journal of Systems Science from Taylor & Francis Journals
Bibliographic data for series maintained by Chris Longhurst ().