Obiekt

Tytuł: Reinforcement learning based iterativa learning control for nonlinear batch process with non-repetitive uncertainty via Koopman operator

Abstract:

To tackle the time and batchwise uncertainty involved in nonlinear batch process, this paper proposes a deep reinforcement learning (DRL) based ILC scheme via Koopman operator. Using the Koopman operator, the original nonlinear system is reformulated into a high-dimensional linear space form. Then, a DRL agent with neural network is introduced into the 2D ILC framework to compensate for non-repetitive uncertainty. ; Meanwhile, the convergence conditions of the proposed ILC scheme are analyzed with a proof through the linear matrix inequality. The proposed method is subsequently applied to a continuous stirring tank reactor (CSTR), and the output of the highdimensional linear model is found to be almost identical to that of the original nonlinear model, with an output error of less than 5%. ; In contrast to the linearized ILC near the equilibrium point, the control performance of the ILC under the high-dimensional model based on Koopman operator is conspicuously augmented. Eventually, the utilisation of deep reinforcement learning to address non-repetitive uncertainty and model mismatch significantly enhances the control performance in comparison to dynamic iterative linearisation and PD-type classical ILC approach.

Opis:

artykuł zamieszczony w: "Journal of Process Control", Vol. 148

Format:

application/pdf

Identyfikator zasobu:

oai:zbc.uz.zgora.pl:94076

DOI:

10.1016/j.jprocont.2025.103402

Strony:

1-11

Jezyk:

eng

Prawa do dysponowania publikacją:

Biblioteka Uniwersytetu Zielonogórskiego

Kolekcje, do których przypisany jest obiekt:

Data ostatniej modyfikacji:

27 mar 2026

Data dodania obiektu:

27 mar 2026

Liczba wyświetleń treści obiektu:

0

Wszystkie dostępne wersje tego obiektu:

https://zbc.uz.zgora.pl/repozytorium/publication/105888

Wyświetl opis w formacie RDF:

RDF

Wyświetl opis w formacie OAI-PMH:

OAI-PMH

Obiekty Podobne

×

Cytowanie

Styl cytowania:

Ta strona wykorzystuje pliki 'cookies'. Więcej informacji