Benutzer: Gast  Login
Titel:

DRLinSPH: an open-source platform using deep reinforcement learning and SPHinXsys for fluid-structure-interaction problems

Dokumenttyp:
Zeitschriftenaufsatz
Autor(en):
Ye, Mai; Ma, Hao; Ren, Yaru; Zhang, Chi; Haidn, Oskar J.; Hu, Xiangyu
Abstract:
Fluid-structure interaction (FSI) problems are characterized by strong nonlinearities arising from complex interactions between fluids and structures. These pose significant challenges for traditional control strategies in optimizing structural motion, often leading to suboptimal performance. In contrast, deep reinforcement learning (DRL), through agent interactions within numerical simulation environments and the approximation of control policies using deep neural networks (DNNs), has shown considerable promise in addressing high-dimensional FSI problems. Furthermore, the training of DRL models necessitates a stable numerical environment, particularly for FSI problems. Smoothed particle hydrodynamics (SPH) offers a flexible and efficient computational approach for modeling large deformations, fractures, and complex interface movements inherent in FSI, outperforming traditional grid-based methods. This work presents DRLinSPH, an open-source Python platform that integrates the SPH-based numerical environment provided by the open-source software SPHinXsys with the mature DRL platform Tianshou to enable parallel training for FSI problems. DRLinSPH has been successfully applied to four FSI scenarios: sloshing suppression using rigid and elastic baffles by controlling displacement or introducing deformation, achieving a maximum wave height reduction of 68.81% and 42.92%, respectively; wave energy harvesting optimization with an 8.25% improvement through an oscillating wave surge converter (OWSC) by regulating the damping characteristics of the Power Take-Off (PTO) system; and muscle-driven fish swimming control in a straight line within vortices. The results demonstrate the platform's accuracy, stability, and scalability, highlighting its potential to advance industrial solutions for complex FSI challenges. © 2025 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
Stichworte:
deep reinforcement learning; fish swimming; fluid-structure interaction; oscillating wave surge converter; sloshing suppression; Smoothed particle hydrodynamics
Dewey Dezimalklassifikation:
620 Ingenieurwissenschaften
Zeitschriftentitel:
Engineering Applications of Computational Fluid Mechanics
Jahr:
2025
Band / Volume:
19
Heft / Issue:
1
Nachgewiesen in:
Scopus
Sprache:
en
Volltext / DOI:
doi:10.1080/19942060.2025.2460677
Verlag / Institution:
Informa UK Limited
E-ISSN:
1994-20601997-003X
Hinweise:
Funding text This work was supported by the China Scholarship Council under Grant [No. 202006120018].
Eingereicht (bei Zeitschrift):
02.10.2024
Angenommen (von Zeitschrift):
02.01.2025
Publikationsdatum:
12.02.2025
TUM Einrichtung:
Lehrstuhl für Aerodynamik und Strömungsmechanik
 BibTeX