Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Pre-trained combustion model and transfer learning in thermoacoustic instability

Qin, Ziyu ; Wang, Xinyao ; Han, Xiao ; Lin, Yuzhen and Zhou, Yuchen LU (2023) In Physics of Fluids 35(3).
Abstract
In this paper, deep learning is involved to comprehend thermoacoustic instability more deeply and achieve early warning more reliably. Flame images and pressure series are acquired in model combustors. A total of seven data domains are obtained by changing the combustor structural parameters. Then, the pre-trained model TIPE (Thermoacoustic Image-Pressure Encoder), containing an image encoder with ResNet architecture and a pressure encoder with transformer architecture, is trained through the contrastive self-supervised task of aligning the image and pressure signals in the embedding space. Furthermore, transfer learning in thermoacoustic instability prediction is performed based on k-nearest neighbors. Results show that the pre-trained... (More)
In this paper, deep learning is involved to comprehend thermoacoustic instability more deeply and achieve early warning more reliably. Flame images and pressure series are acquired in model combustors. A total of seven data domains are obtained by changing the combustor structural parameters. Then, the pre-trained model TIPE (Thermoacoustic Image-Pressure Encoder), containing an image encoder with ResNet architecture and a pressure encoder with transformer architecture, is trained through the contrastive self-supervised task of aligning the image and pressure signals in the embedding space. Furthermore, transfer learning in thermoacoustic instability prediction is performed based on k-nearest neighbors. Results show that the pre-trained model can better resist the negative effect caused by class imbalance. The weighted F1 score of the pre-trained model is 6.72% and 2.61% larger than supervised models in zero-shot transfer and few-shot transfer, respectively. It is inferred that the more generic features encoded by TIPE result in superior generalization in comparison with traditional supervised methods. Moreover, our proposed method is insensitive to the thresholds of determining thermoacoustic states. Principal component analysis reveals the physical interpretability preliminarily through the connection between feature principal components and pressure fluctuation amplitudes. Finally, the key spatial region of flame images and temporal interval of pressure series are visualized by class activation map and global attention scores. (Less)
Please use this url to cite or link to this publication:
author
; ; ; and
publishing date
type
Contribution to journal
publication status
published
in
Physics of Fluids
volume
35
issue
3
article number
037117
publisher
American Institute of Physics (AIP)
external identifiers
  • scopus:85150168775
ISSN
1070-6631
DOI
10.1063/5.0142378
language
English
LU publication?
no
id
bf63787f-e24b-4a46-a09c-3385796bb783
date added to LUP
2025-09-12 20:59:25
date last changed
2025-09-30 13:06:53
@article{bf63787f-e24b-4a46-a09c-3385796bb783,
  abstract     = {{In this paper, deep learning is involved to comprehend thermoacoustic instability more deeply and achieve early warning more reliably. Flame images and pressure series are acquired in model combustors. A total of seven data domains are obtained by changing the combustor structural parameters. Then, the pre-trained model TIPE (Thermoacoustic Image-Pressure Encoder), containing an image encoder with ResNet architecture and a pressure encoder with transformer architecture, is trained through the contrastive self-supervised task of aligning the image and pressure signals in the embedding space. Furthermore, transfer learning in thermoacoustic instability prediction is performed based on k-nearest neighbors. Results show that the pre-trained model can better resist the negative effect caused by class imbalance. The weighted F1 score of the pre-trained model is 6.72% and 2.61% larger than supervised models in zero-shot transfer and few-shot transfer, respectively. It is inferred that the more generic features encoded by TIPE result in superior generalization in comparison with traditional supervised methods. Moreover, our proposed method is insensitive to the thresholds of determining thermoacoustic states. Principal component analysis reveals the physical interpretability preliminarily through the connection between feature principal components and pressure fluctuation amplitudes. Finally, the key spatial region of flame images and temporal interval of pressure series are visualized by class activation map and global attention scores.}},
  author       = {{Qin, Ziyu and Wang, Xinyao and Han, Xiao and Lin, Yuzhen and Zhou, Yuchen}},
  issn         = {{1070-6631}},
  language     = {{eng}},
  month        = {{03}},
  number       = {{3}},
  publisher    = {{American Institute of Physics (AIP)}},
  series       = {{Physics of Fluids}},
  title        = {{Pre-trained combustion model and transfer learning in thermoacoustic instability}},
  url          = {{http://dx.doi.org/10.1063/5.0142378}},
  doi          = {{10.1063/5.0142378}},
  volume       = {{35}},
  year         = {{2023}},
}