Data-driven reduced-order models via regularised Operator Inference for a single-injector combustion process

This paper derives predictive reduced-order models for rocket engine combustion dynamics via Operator Inference, a scientific machine learning approach that blends data-driven learning with physics-based modelling. The non-intrusive nature of the approach enables variable transformations that expose...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Journal of the Royal Society of New Zealand Jg. 51; H. 2; S. 194 - 211
Hauptverfasser: McQuarrie, Shane A., Huang, Cheng, Willcox, Karen E.
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Wellington Taylor & Francis 01.06.2021
Taylor & Francis Ltd
Schlagworte:
Accuracy
Algorithms
Combustion
Computational efficiency
Computer applications
Computing costs
data-driven model
data-driven reduced model
Dynamic stability
Dynamics
Inference
Injectors
Learning algorithms
Machine learning
Mathematical models
MATHEMATICS AND COMPUTING
Model accuracy
Model reduction
Operator Inference
Optimization
Physics
Reduced order models
Regularization
Rocket engines
scientific machine learning
Size reduction of materials
Time integration
Training
ISSN:0303-6758, 1175-8899, 1175-8899
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:This paper derives predictive reduced-order models for rocket engine combustion dynamics via Operator Inference, a scientific machine learning approach that blends data-driven learning with physics-based modelling. The non-intrusive nature of the approach enables variable transformations that expose system structure. The specific contribution of this paper is to advance the formulation robustness and algorithmic scalability of the Operator Inference approach. Regularisation is introduced to the formulation to avoid over-fitting. The task of determining an optimal regularisation is posed as an optimisation problem that balances training error and stability of long-time integration dynamics. A scalable algorithm and open-source implementation are presented, then demonstrated for a single-injector rocket combustion example. This example exhibits rich dynamics that are difficult to capture with state-of-the-art reduced models. With appropriate regularisation and an informed selection of learning variables, the reduced-order models exhibit high accuracy in re-predicting the training regime and acceptable accuracy in predicting future dynamics, while achieving close to a million times speedup in computational cost. When compared to a state-of-the-art model reduction method, the Operator Inference models provide the same or better accuracy at approximately one thousandth of the computational cost.
Bibliographie:Includes graphs, illustrations, notes, references, table
Archived by the National Library of New Zealand
ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
SC0019303; FA9550-17-1-0195
USDOE Office of Science (SC)
US Air Force Office of Scientific Research (AFOSR)
ISSN:0303-6758
1175-8899
1175-8899
DOI:10.1080/03036758.2020.1863237