Early detection of combustion instability from hi-speed flame images via deep learning and symbolic time series analysis

Soumalya Sarkar, Kin G. Lore, Soumik Sarkar, Vikram Ramanan, Satyanarayanan R. Chakravarthy, Shashi Phoha, Asok Ray

Research output: Chapter in Book/Report/Conference proceedingConference contribution

11 Citations (Scopus)

Abstract

Combustion instability, characterized by self-sustained, large-amplitude pressure oscillations and periodic shedding of coherent vortex structures at varied spatial scales, has many detrimental effects on flight-propulsion dynamics and structural integrity of gas turbine engines. Hence, its early detection is one of the important tasks in engine health monitoring and prognostics. This paper proposes a dynamic data-driven approach, where a large volume of sequential hi-speed (greyscale) images is used to analyze the temporal evolution of coherent structures in combustion chamber for early detection of combustion instability at various operating conditions. The proposed hierarchical approach involves extracting low-dimensional semantic features from images using Deep Neural Networks followed by capturing the temporal evolution of the extracted features using Symbolic Time Series Analysis (STSA). Extensive experimental data have been collected in a swirl-stabilized dump combustor at various operating conditions for validation of the proposed approach. Intermediate layer visualization of deep learning reveals that meaningful shape-features from the flame images are extracted, which enables the temporal modeling layer to enhance the class separability between stable and unstable regions. At the same time, the semantic nature of intermediate features enables expert-guided data exploration that can lead to better understanding of the underlying physics. To the best of the authors knowledge, this paper presents one of the early applications of the recently reported Deep Learning tools in the area of prognostics and health management (PHM).

Original languageEnglish (US)
Title of host publicationPHM 2015 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2015
EditorsMatthew J. Daigle, Anibal Bregon
PublisherPrognostics and Health Management Society
Pages353-362
Number of pages10
ISBN (Electronic)9781936263202
StatePublished - Jan 1 2015
Event2015 Annual Conference of the Prognostics and Health Management Society, PHM 2015 - San Diego, United States
Duration: Oct 18 2015Oct 22 2015

Publication series

NameProceedings of the Annual Conference of the Prognostics and Health Management Society, PHM
ISSN (Print)2325-0178

Other

Other2015 Annual Conference of the Prognostics and Health Management Society, PHM 2015
CountryUnited States
CitySan Diego
Period10/18/1510/22/15

Fingerprint

Time series analysis
Semantics
Health
Learning
Structural integrity
Combustion chambers
Combustors
Propulsion
Gas turbines
Vortex flow
Turbines
Visualization
Physics
Engines
Monitoring
Gases
Pressure
Deep learning
Deep neural networks

All Science Journal Classification (ASJC) codes

  • Information Systems
  • Electrical and Electronic Engineering
  • Health Information Management
  • Computer Science Applications

Cite this

Sarkar, S., Lore, K. G., Sarkar, S., Ramanan, V., Chakravarthy, S. R., Phoha, S., & Ray, A. (2015). Early detection of combustion instability from hi-speed flame images via deep learning and symbolic time series analysis. In M. J. Daigle, & A. Bregon (Eds.), PHM 2015 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2015 (pp. 353-362). (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM). Prognostics and Health Management Society.
Sarkar, Soumalya ; Lore, Kin G. ; Sarkar, Soumik ; Ramanan, Vikram ; Chakravarthy, Satyanarayanan R. ; Phoha, Shashi ; Ray, Asok. / Early detection of combustion instability from hi-speed flame images via deep learning and symbolic time series analysis. PHM 2015 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2015. editor / Matthew J. Daigle ; Anibal Bregon. Prognostics and Health Management Society, 2015. pp. 353-362 (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM).
@inproceedings{eaf27a510c3a442ca5eecf081d3a685b,
title = "Early detection of combustion instability from hi-speed flame images via deep learning and symbolic time series analysis",
abstract = "Combustion instability, characterized by self-sustained, large-amplitude pressure oscillations and periodic shedding of coherent vortex structures at varied spatial scales, has many detrimental effects on flight-propulsion dynamics and structural integrity of gas turbine engines. Hence, its early detection is one of the important tasks in engine health monitoring and prognostics. This paper proposes a dynamic data-driven approach, where a large volume of sequential hi-speed (greyscale) images is used to analyze the temporal evolution of coherent structures in combustion chamber for early detection of combustion instability at various operating conditions. The proposed hierarchical approach involves extracting low-dimensional semantic features from images using Deep Neural Networks followed by capturing the temporal evolution of the extracted features using Symbolic Time Series Analysis (STSA). Extensive experimental data have been collected in a swirl-stabilized dump combustor at various operating conditions for validation of the proposed approach. Intermediate layer visualization of deep learning reveals that meaningful shape-features from the flame images are extracted, which enables the temporal modeling layer to enhance the class separability between stable and unstable regions. At the same time, the semantic nature of intermediate features enables expert-guided data exploration that can lead to better understanding of the underlying physics. To the best of the authors knowledge, this paper presents one of the early applications of the recently reported Deep Learning tools in the area of prognostics and health management (PHM).",
author = "Soumalya Sarkar and Lore, {Kin G.} and Soumik Sarkar and Vikram Ramanan and Chakravarthy, {Satyanarayanan R.} and Shashi Phoha and Asok Ray",
year = "2015",
month = "1",
day = "1",
language = "English (US)",
series = "Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM",
publisher = "Prognostics and Health Management Society",
pages = "353--362",
editor = "Daigle, {Matthew J.} and Anibal Bregon",
booktitle = "PHM 2015 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2015",

}

Sarkar, S, Lore, KG, Sarkar, S, Ramanan, V, Chakravarthy, SR, Phoha, S & Ray, A 2015, Early detection of combustion instability from hi-speed flame images via deep learning and symbolic time series analysis. in MJ Daigle & A Bregon (eds), PHM 2015 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2015. Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM, Prognostics and Health Management Society, pp. 353-362, 2015 Annual Conference of the Prognostics and Health Management Society, PHM 2015, San Diego, United States, 10/18/15.

Early detection of combustion instability from hi-speed flame images via deep learning and symbolic time series analysis. / Sarkar, Soumalya; Lore, Kin G.; Sarkar, Soumik; Ramanan, Vikram; Chakravarthy, Satyanarayanan R.; Phoha, Shashi; Ray, Asok.

PHM 2015 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2015. ed. / Matthew J. Daigle; Anibal Bregon. Prognostics and Health Management Society, 2015. p. 353-362 (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Early detection of combustion instability from hi-speed flame images via deep learning and symbolic time series analysis

AU - Sarkar, Soumalya

AU - Lore, Kin G.

AU - Sarkar, Soumik

AU - Ramanan, Vikram

AU - Chakravarthy, Satyanarayanan R.

AU - Phoha, Shashi

AU - Ray, Asok

PY - 2015/1/1

Y1 - 2015/1/1

N2 - Combustion instability, characterized by self-sustained, large-amplitude pressure oscillations and periodic shedding of coherent vortex structures at varied spatial scales, has many detrimental effects on flight-propulsion dynamics and structural integrity of gas turbine engines. Hence, its early detection is one of the important tasks in engine health monitoring and prognostics. This paper proposes a dynamic data-driven approach, where a large volume of sequential hi-speed (greyscale) images is used to analyze the temporal evolution of coherent structures in combustion chamber for early detection of combustion instability at various operating conditions. The proposed hierarchical approach involves extracting low-dimensional semantic features from images using Deep Neural Networks followed by capturing the temporal evolution of the extracted features using Symbolic Time Series Analysis (STSA). Extensive experimental data have been collected in a swirl-stabilized dump combustor at various operating conditions for validation of the proposed approach. Intermediate layer visualization of deep learning reveals that meaningful shape-features from the flame images are extracted, which enables the temporal modeling layer to enhance the class separability between stable and unstable regions. At the same time, the semantic nature of intermediate features enables expert-guided data exploration that can lead to better understanding of the underlying physics. To the best of the authors knowledge, this paper presents one of the early applications of the recently reported Deep Learning tools in the area of prognostics and health management (PHM).

AB - Combustion instability, characterized by self-sustained, large-amplitude pressure oscillations and periodic shedding of coherent vortex structures at varied spatial scales, has many detrimental effects on flight-propulsion dynamics and structural integrity of gas turbine engines. Hence, its early detection is one of the important tasks in engine health monitoring and prognostics. This paper proposes a dynamic data-driven approach, where a large volume of sequential hi-speed (greyscale) images is used to analyze the temporal evolution of coherent structures in combustion chamber for early detection of combustion instability at various operating conditions. The proposed hierarchical approach involves extracting low-dimensional semantic features from images using Deep Neural Networks followed by capturing the temporal evolution of the extracted features using Symbolic Time Series Analysis (STSA). Extensive experimental data have been collected in a swirl-stabilized dump combustor at various operating conditions for validation of the proposed approach. Intermediate layer visualization of deep learning reveals that meaningful shape-features from the flame images are extracted, which enables the temporal modeling layer to enhance the class separability between stable and unstable regions. At the same time, the semantic nature of intermediate features enables expert-guided data exploration that can lead to better understanding of the underlying physics. To the best of the authors knowledge, this paper presents one of the early applications of the recently reported Deep Learning tools in the area of prognostics and health management (PHM).

UR - http://www.scopus.com/inward/record.url?scp=85016118286&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85016118286&partnerID=8YFLogxK

M3 - Conference contribution

T3 - Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM

SP - 353

EP - 362

BT - PHM 2015 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2015

A2 - Daigle, Matthew J.

A2 - Bregon, Anibal

PB - Prognostics and Health Management Society

ER -

Sarkar S, Lore KG, Sarkar S, Ramanan V, Chakravarthy SR, Phoha S et al. Early detection of combustion instability from hi-speed flame images via deep learning and symbolic time series analysis. In Daigle MJ, Bregon A, editors, PHM 2015 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2015. Prognostics and Health Management Society. 2015. p. 353-362. (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM).