Early Detection of High-Frequency Combustion Instability Using Deep Learning Model

Gyeonghyun Han; Kyu Tae Kim

doi:10.15231/jksc.2026.31.2.010

All Issue

2026 Vol.31, Issue 2 Preview Page Next Page

Research Article

Early Detection of High-Frequency Combustion Instability Using Deep Learning Model 딥러닝 모델을 이용한 고주파 연소 불안정 조기 감지

30 June 2026. pp. 10-19

PDF XML

Abstract

This study examines the feasibility of real-time early detection of combustion instability using a ResNet-18 model trained on recurrence plot images derived from dynamic pressure signals. Experiments were conducted in a hydrogen-fueled premixed combustor by gradually increasing the fuel flow rate from three stable initial velocities, which induced several high-frequency instability modes. A model trained on a single mode could detect instability onset, but its performance depended on the frequency difference between the trained and target instability modes. A combined-frequency model showed effective detection performance, although improper frequency selection caused overly sensitive responses in stable regions. These results indicate that robust early detection requires not only multi-band training, but also careful selection of representative instability frequencies.

Keywords

Combustion instability

Hydrogen

Early detection

Deep learning

Recurrence plot

References

T. Asai, S. Dodo, M. Karishuku, N. Yagi, Y. Akiyama, A. Hayashi, Performance of multiple-injection dry low-NOx combustors on hydrogen-rich syngas fuel in an IGCC pilot plant, J. Eng. Gas Turbines Power 137(9) (2015) 091504.

S.M. Correa, A review of NOx formation under gas-turbine combustion conditions, Combust. Sci. Technol. 87 (1993) 329-362.

10.1080/00102209208947221

H.H.W. Funke, N. Beckmann, S. Abanteriba, An overview on dry low NOx micromix combustor development for hydrogen-rich gas turbine applications, Int. J. Hydrog. Energy 44(13) (2019) 6978-6990.

10.1016/j.ijhydene.2019.01.161

L. Angello, B. Noble, R. Steele, M. Cohen, B. Emerson, Dry low NOx (DLN) combustion system operability considerations, ASME Turbo Expo (2022), paper GT2022-82667.

10.1115/GT2022-82667

T. Lieuwen, B.T. Zinn, The role of equivalence ratio oscillations in driving combustion instabilities in low NOx gas turbines, Symp. (Int.) Combust. 27 (1998) 1809-1816.

10.1016/S0082-0784(98)80022-2

D. Kim, Review on the development trend of hydrogen gas turbine combustion technology, J. Korean Soc. Combust. 24(4) (2019) 1-10.

10.15231/jksc.2019.24.4.001

J. Jung, D. Kim, M.K. Kim, J. Hwang, W.J. Lee, Analysis of combustion instability characteristics of a single-nozzle combustor in a natural gas-hydrogen co-firing gas turbine under varying operating pressures, J. Korean Soc. Combust. 30(2) (2025) 27-35.

10.15231/jksc.2025.30.2.027

H. Paniez, S. Marragou, H. Magnes, T. Schuller, High-frequency thermo-acoustic instability in a dual swirl H2 burner, Proc. Combust. Inst. 40 (2024) 105679.

10.1016/j.proci.2024.105679

H. Kang, K.T. Kim, Combustion dynamics of multi-element lean-premixed hydrogen-air flame ensemble, Combust. Flame 233 (2021) 111585.

10.1016/j.combustflame.2021.111585

J.G. Aguilar, E. Æsøy, J.R. Dawson, The influence of hydrogen on the stability of a perfectly premixed combustor, Combust. Flame 245 (2022) 112323.

10.1016/j.combustflame.2022.112323

S. Choi, D. Kim, Diagnosis of combustion instability using waveform pattern of dynamic pressure signal, J. Korean Soc. Combust. 29(1) (2024) 33-38.

10.15231/jksc.2024.29.1.033

D.J. Jang, S. Joo, M.K. Kim, J. Hwang, M.C. Lee, Novel combustion instability diagnosis method in a hydrogen/natural gas co-firing gas turbine combustor using a combination of four criteria: Temporal kurtosis, permutation entropy, energy of entropy, and zero-crossing rate, Int. J. Hydrog. Energy 85 (2024) 773-782.

T. Hashimoto, H. Shibuya, H. Gotoda, Y. Ohmichi, S. Matsuyama, Spatiotemporal dynamics and early detection of thermoacoustic combustion instability in a model rocket combustor, Phys. Rev. E 99(3) (2019) 032208.

10.1103/PhysRevE.99.032208

R. Meloni, N. Giannini, Real time precursor calculation for the early detection of combustion instabilities, J. Eng. Gas Turbines Power 144(9) (2022), 091016.

A. Cellier, C.J. Lapeyre, G. Öztarlik, T. Poinsot, T. Schuller, L. Selle, Detection of precursors of combustion instability using convolutional recurrent neural networks, Combust. Flame 233 (2021) 111558.

10.1016/j.combustflame.2021.111558

S. Sarkar, K.G. Lore, S. Sarkar, V. Ramanan, S.R. Chakravarthy, S. Phoha, A. Ray, Early detection of combustion instability from hi-speed flame images via deep learning and symbolic time series analysis, Annual Conference of the PHM Society (2015), pp. 353-362.

10.36001/phmconf.2015.v7i1.2723

G. Xu, B. Wang, Y. Guan, Z. Wang, P. Liu, Early detection of thermoacoustic instability in a solid rocket motor: A generative adversarial network approach with limited data, Appl. Energy 373 (2024) 123776.

10.1016/j.apenergy.2024.123776

T.M. Bury, R.I. Sujith, I. Pavithran, M. Scheffer, T.M. Lenton, M. Anand, C.T. Bauch, Deep learning for early warning signals of tipping points, Proc. Natl. Acad. Sci. U.S.A., 118(39) (2021) e2106140118.

10.1073/pnas.210614011834544867PMC8488604

D. Park, K.T. Kim, Rotational symmetry-driven modal dynamics of high-frequency transverse instabilities in a lean-premixed multislit hydrogen combustor, Combust. Flame 245 (2022) 112356.

10.1016/j.combustflame.2022.112356

J.P. Eckmann, S.O. Kamphorst, D. Ruelle, Recurrence plots of dynamical systems, Europhys. Lett. 4(9) (1987) 973-977.

10.1209/0295-5075/4/9/004

N. Marwan, M.C. Romano, M. Thiel, J. Kurths, Recurrence plots for the analysis of complex systems, Phys. Rep. 438 (2007) 237-329.

10.1016/j.physrep.2006.11.001

K. He, X. Zhang, S. Ren, J. Sun, Deep residual learning for image recognition, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, 770-778.

J.F. Bourgouin, D. Durox, J.P. Moeck, T. Schuller, S. Candel, Self-sustained instabilities in an annular combustor coupled by azimuthal and longitudinal acoustic modes, ASME Turbo Expo (2013), paper GT2013-95010.

10.1115/GT2013-95010

Information

Publisher :The Korean Society of Combustion
Publisher(Ko) :한국연소학회
Journal Title :Journal of the Korean Society of Combustion
Journal Title(Ko) :한국연소학회지
Volume : 31
No :2
Pages :10-19
Received Date : 2026-03-26
Revised Date : 2026-04-21
Accepted Date : 2026-04-21
DOI :https://doi.org/10.15231/jksc.2026.31.2.010

Journal of the Korean Society of Combustion ISSN:1226-0959(Print) 2466-2089(Online) 한국연소학회지

All Issue