TY - JOUR
T1 - Impact of sensor placement on mode observability and LQG control of a thermoacoustic system
AU - Chen, Xiaoling
AU - Fathy, Hosam
AU - O'Connor, Jacqueline
N1 - Funding Information:
This material is based upon work supported by the National Science Foundation under Grant No.e ∆MMI-1728307. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the ffiiews of the National Science Foundation.
Publisher Copyright:
Copyright © 2020 The Authors.
PY - 2020
Y1 - 2020
N2 - This paper investigates the effect of sensor placement on the observability and LQG control of a thermoacoustic model. This model describes combustion instability in a one-dimensional combustor, called a Rijke tube. The transfer function describing this model is transcendental because of the time delay terms in the heat release dynamics. We apply Padé approximation to achieve a finite-dimensional transfer function and truncate the system by neglecting states with low Hankel singular values. We then analyze the impact of the placement and number of sensors on the observability of each mode of the resulting reduced-order model. Next, we design an LQG controller for suppressing pressure oscillations in the simplified thermoacoustic system. We find that placing sensors near the model's pressure nodes slows down the rate at which LQG control attenuates pressure oscillations, increases the control effort required for this attenuation, and worsens the controller's robustness.
AB - This paper investigates the effect of sensor placement on the observability and LQG control of a thermoacoustic model. This model describes combustion instability in a one-dimensional combustor, called a Rijke tube. The transfer function describing this model is transcendental because of the time delay terms in the heat release dynamics. We apply Padé approximation to achieve a finite-dimensional transfer function and truncate the system by neglecting states with low Hankel singular values. We then analyze the impact of the placement and number of sensors on the observability of each mode of the resulting reduced-order model. Next, we design an LQG controller for suppressing pressure oscillations in the simplified thermoacoustic system. We find that placing sensors near the model's pressure nodes slows down the rate at which LQG control attenuates pressure oscillations, increases the control effort required for this attenuation, and worsens the controller's robustness.
UR - http://www.scopus.com/inward/record.url?scp=85107811983&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85107811983&partnerID=8YFLogxK
U2 - 10.1016/j.ifacol.2020.12.2466
DO - 10.1016/j.ifacol.2020.12.2466
M3 - Conference article
AN - SCOPUS:85107811983
SN - 2405-8963
VL - 53
SP - 4214
EP - 4221
JO - IFAC-PapersOnLine
JF - IFAC-PapersOnLine
T2 - 21st IFAC World Congress 2020
Y2 - 12 July 2020 through 17 July 2020
ER -