Early detection of lean blow out (LBO) via generalized D-Markov machine construction

Soumalya Sarkar, Asok Ray, Achintya Mukhopadhyay, Rajendra R. Chaudhari, Swarnendu Sen

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

6 Citations (Scopus)

Abstract

This paper develops a method for early detection of lean-blow-out (LBO) in combustion systems by extracting low-dimensional features from chemiluminescence time series of optical sensor data. In the proposed method, symbol strings are generated by partitioning the (finite-length) time series to construct a special class of probabilistic finite state automata (PFSA), called D-Markov machines. These PFSA have a deterministic algebraic structure and their states are represented by symbol blocks of length D or less. The states of D-Markov machines are constructed in two steps: (i) state splitting, i.e., the states are split based on their information contents, and (ii) state merging, i.e., two or more states (of possibly different lengths) are merged together to form a new state without any significant loss of their embedded information. The modeling complexity (i.e., the number of states) of a D-Markov machine is observed to be drastically reduced as the combustion system approaches LBO. The prediction of LBO is posed as a pattern classification problem based on different ranges of equivalence ratio of the flame. It is shown that, over a wide range of air-fuel premixing, a generalized D-Markov machine (i.e., with D > 1) performs better than a D-Markov machine with D = 1 as a predictor of LBO.

Original languageEnglish (US)
Title of host publication2014 American Control Conference, ACC 2014
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages3041-3046
Number of pages6
ISBN (Print)9781479932726
DOIs
StatePublished - Jan 1 2014
Event2014 American Control Conference, ACC 2014 - Portland, OR, United States
Duration: Jun 4 2014Jun 6 2014

Publication series

NameProceedings of the American Control Conference
ISSN (Print)0743-1619

Other

Other2014 American Control Conference, ACC 2014
CountryUnited States
CityPortland, OR
Period6/4/146/6/14

Fingerprint

Finite automata
Time series
Chemiluminescence
Optical sensors
Merging
Pattern recognition
Air

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Cite this

Sarkar, S., Ray, A., Mukhopadhyay, A., Chaudhari, R. R., & Sen, S. (2014). Early detection of lean blow out (LBO) via generalized D-Markov machine construction. In 2014 American Control Conference, ACC 2014 (pp. 3041-3046). [6859048] (Proceedings of the American Control Conference). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ACC.2014.6859048
Sarkar, Soumalya ; Ray, Asok ; Mukhopadhyay, Achintya ; Chaudhari, Rajendra R. ; Sen, Swarnendu. / Early detection of lean blow out (LBO) via generalized D-Markov machine construction. 2014 American Control Conference, ACC 2014. Institute of Electrical and Electronics Engineers Inc., 2014. pp. 3041-3046 (Proceedings of the American Control Conference).
@inproceedings{142510c34f5d406e8a80de6f4a269859,
title = "Early detection of lean blow out (LBO) via generalized D-Markov machine construction",
abstract = "This paper develops a method for early detection of lean-blow-out (LBO) in combustion systems by extracting low-dimensional features from chemiluminescence time series of optical sensor data. In the proposed method, symbol strings are generated by partitioning the (finite-length) time series to construct a special class of probabilistic finite state automata (PFSA), called D-Markov machines. These PFSA have a deterministic algebraic structure and their states are represented by symbol blocks of length D or less. The states of D-Markov machines are constructed in two steps: (i) state splitting, i.e., the states are split based on their information contents, and (ii) state merging, i.e., two or more states (of possibly different lengths) are merged together to form a new state without any significant loss of their embedded information. The modeling complexity (i.e., the number of states) of a D-Markov machine is observed to be drastically reduced as the combustion system approaches LBO. The prediction of LBO is posed as a pattern classification problem based on different ranges of equivalence ratio of the flame. It is shown that, over a wide range of air-fuel premixing, a generalized D-Markov machine (i.e., with D > 1) performs better than a D-Markov machine with D = 1 as a predictor of LBO.",
author = "Soumalya Sarkar and Asok Ray and Achintya Mukhopadhyay and Chaudhari, {Rajendra R.} and Swarnendu Sen",
year = "2014",
month = "1",
day = "1",
doi = "10.1109/ACC.2014.6859048",
language = "English (US)",
isbn = "9781479932726",
series = "Proceedings of the American Control Conference",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "3041--3046",
booktitle = "2014 American Control Conference, ACC 2014",
address = "United States",

}

Sarkar, S, Ray, A, Mukhopadhyay, A, Chaudhari, RR & Sen, S 2014, Early detection of lean blow out (LBO) via generalized D-Markov machine construction. in 2014 American Control Conference, ACC 2014., 6859048, Proceedings of the American Control Conference, Institute of Electrical and Electronics Engineers Inc., pp. 3041-3046, 2014 American Control Conference, ACC 2014, Portland, OR, United States, 6/4/14. https://doi.org/10.1109/ACC.2014.6859048

Early detection of lean blow out (LBO) via generalized D-Markov machine construction. / Sarkar, Soumalya; Ray, Asok; Mukhopadhyay, Achintya; Chaudhari, Rajendra R.; Sen, Swarnendu.

2014 American Control Conference, ACC 2014. Institute of Electrical and Electronics Engineers Inc., 2014. p. 3041-3046 6859048 (Proceedings of the American Control Conference).

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

TY - GEN

T1 - Early detection of lean blow out (LBO) via generalized D-Markov machine construction

AU - Sarkar, Soumalya

AU - Ray, Asok

AU - Mukhopadhyay, Achintya

AU - Chaudhari, Rajendra R.

AU - Sen, Swarnendu

PY - 2014/1/1

Y1 - 2014/1/1

N2 - This paper develops a method for early detection of lean-blow-out (LBO) in combustion systems by extracting low-dimensional features from chemiluminescence time series of optical sensor data. In the proposed method, symbol strings are generated by partitioning the (finite-length) time series to construct a special class of probabilistic finite state automata (PFSA), called D-Markov machines. These PFSA have a deterministic algebraic structure and their states are represented by symbol blocks of length D or less. The states of D-Markov machines are constructed in two steps: (i) state splitting, i.e., the states are split based on their information contents, and (ii) state merging, i.e., two or more states (of possibly different lengths) are merged together to form a new state without any significant loss of their embedded information. The modeling complexity (i.e., the number of states) of a D-Markov machine is observed to be drastically reduced as the combustion system approaches LBO. The prediction of LBO is posed as a pattern classification problem based on different ranges of equivalence ratio of the flame. It is shown that, over a wide range of air-fuel premixing, a generalized D-Markov machine (i.e., with D > 1) performs better than a D-Markov machine with D = 1 as a predictor of LBO.

AB - This paper develops a method for early detection of lean-blow-out (LBO) in combustion systems by extracting low-dimensional features from chemiluminescence time series of optical sensor data. In the proposed method, symbol strings are generated by partitioning the (finite-length) time series to construct a special class of probabilistic finite state automata (PFSA), called D-Markov machines. These PFSA have a deterministic algebraic structure and their states are represented by symbol blocks of length D or less. The states of D-Markov machines are constructed in two steps: (i) state splitting, i.e., the states are split based on their information contents, and (ii) state merging, i.e., two or more states (of possibly different lengths) are merged together to form a new state without any significant loss of their embedded information. The modeling complexity (i.e., the number of states) of a D-Markov machine is observed to be drastically reduced as the combustion system approaches LBO. The prediction of LBO is posed as a pattern classification problem based on different ranges of equivalence ratio of the flame. It is shown that, over a wide range of air-fuel premixing, a generalized D-Markov machine (i.e., with D > 1) performs better than a D-Markov machine with D = 1 as a predictor of LBO.

UR - http://www.scopus.com/inward/record.url?scp=84905695671&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84905695671&partnerID=8YFLogxK

U2 - 10.1109/ACC.2014.6859048

DO - 10.1109/ACC.2014.6859048

M3 - Conference contribution

AN - SCOPUS:84905695671

SN - 9781479932726

T3 - Proceedings of the American Control Conference

SP - 3041

EP - 3046

BT - 2014 American Control Conference, ACC 2014

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Sarkar S, Ray A, Mukhopadhyay A, Chaudhari RR, Sen S. Early detection of lean blow out (LBO) via generalized D-Markov machine construction. In 2014 American Control Conference, ACC 2014. Institute of Electrical and Electronics Engineers Inc. 2014. p. 3041-3046. 6859048. (Proceedings of the American Control Conference). https://doi.org/10.1109/ACC.2014.6859048