Stochastic estimation using modern methods in machine learning

Andrew S. Tenney, Mark N Glauser, Zachary P. Berger

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

Abstract

In many areas of data science, Deep Neural Networks have exhibited a remarkable ability to learn complex, non-linear relationships between sets of variables. In this paper, we apply this network architecture to the task of stochastic estimation, first proposed by Adrian over 40 years ago. Two Deep Neural Networks (DNNs) are trained to estimate the pressure at selected locations in azimuthal and stream-wise arrays of pressure transducers situated just outside a Mach 0.6 jet, given instantaneous pressure measurements made at other locations in the arrays. The estimated pressure is compared with the instantaneous pressure fluctuations measured at each of the selected locations, as well as the estimates made using traditional Linear Stochastic Estimation (LSE). The root-mean-square error between the values predicted by the DNN and those measured by the transducers is shown to be nearly identical to the error associated with the traditional LSE method in the case of the azimuthal array. For the stream-wise configuration, the DNN shows an approximately 60% reduction in error over the LSE model. In addition, some limitations and possible extensions of this method are discussed.

Original languageEnglish (US)
Title of host publicationAIAA Scitech 2019 Forum
PublisherAmerican Institute of Aeronautics and Astronautics Inc, AIAA
ISBN (Print)9781624105784
DOIs
StatePublished - Jan 1 2019
EventAIAA Scitech Forum, 2019 - San Diego, United States
Duration: Jan 7 2019Jan 11 2019

Publication series

NameAIAA Scitech 2019 Forum

Conference

ConferenceAIAA Scitech Forum, 2019
CountryUnited States
CitySan Diego
Period1/7/191/11/19

Fingerprint

Learning systems
Pressure transducers
Pressure measurement
Network architecture
Mean square error
Mach number
Transducers
Deep neural networks

ASJC Scopus subject areas

  • Aerospace Engineering

Cite this

Tenney, A. S., Glauser, M. N., & Berger, Z. P. (2019). Stochastic estimation using modern methods in machine learning. In AIAA Scitech 2019 Forum (AIAA Scitech 2019 Forum). American Institute of Aeronautics and Astronautics Inc, AIAA. https://doi.org/10.2514/6.2019-0324

Stochastic estimation using modern methods in machine learning. / Tenney, Andrew S.; Glauser, Mark N; Berger, Zachary P.

AIAA Scitech 2019 Forum. American Institute of Aeronautics and Astronautics Inc, AIAA, 2019. (AIAA Scitech 2019 Forum).

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

Tenney, AS, Glauser, MN & Berger, ZP 2019, Stochastic estimation using modern methods in machine learning. in AIAA Scitech 2019 Forum. AIAA Scitech 2019 Forum, American Institute of Aeronautics and Astronautics Inc, AIAA, AIAA Scitech Forum, 2019, San Diego, United States, 1/7/19. https://doi.org/10.2514/6.2019-0324
Tenney AS, Glauser MN, Berger ZP. Stochastic estimation using modern methods in machine learning. In AIAA Scitech 2019 Forum. American Institute of Aeronautics and Astronautics Inc, AIAA. 2019. (AIAA Scitech 2019 Forum). https://doi.org/10.2514/6.2019-0324
Tenney, Andrew S. ; Glauser, Mark N ; Berger, Zachary P. / Stochastic estimation using modern methods in machine learning. AIAA Scitech 2019 Forum. American Institute of Aeronautics and Astronautics Inc, AIAA, 2019. (AIAA Scitech 2019 Forum).
@inproceedings{260ab3f430fa492d8b30232cf1d270ba,
title = "Stochastic estimation using modern methods in machine learning",
abstract = "In many areas of data science, Deep Neural Networks have exhibited a remarkable ability to learn complex, non-linear relationships between sets of variables. In this paper, we apply this network architecture to the task of stochastic estimation, first proposed by Adrian over 40 years ago. Two Deep Neural Networks (DNNs) are trained to estimate the pressure at selected locations in azimuthal and stream-wise arrays of pressure transducers situated just outside a Mach 0.6 jet, given instantaneous pressure measurements made at other locations in the arrays. The estimated pressure is compared with the instantaneous pressure fluctuations measured at each of the selected locations, as well as the estimates made using traditional Linear Stochastic Estimation (LSE). The root-mean-square error between the values predicted by the DNN and those measured by the transducers is shown to be nearly identical to the error associated with the traditional LSE method in the case of the azimuthal array. For the stream-wise configuration, the DNN shows an approximately 60{\%} reduction in error over the LSE model. In addition, some limitations and possible extensions of this method are discussed.",
author = "Tenney, {Andrew S.} and Glauser, {Mark N} and Berger, {Zachary P.}",
year = "2019",
month = "1",
day = "1",
doi = "10.2514/6.2019-0324",
language = "English (US)",
isbn = "9781624105784",
series = "AIAA Scitech 2019 Forum",
publisher = "American Institute of Aeronautics and Astronautics Inc, AIAA",
booktitle = "AIAA Scitech 2019 Forum",

}

TY - GEN

T1 - Stochastic estimation using modern methods in machine learning

AU - Tenney, Andrew S.

AU - Glauser, Mark N

AU - Berger, Zachary P.

PY - 2019/1/1

Y1 - 2019/1/1

N2 - In many areas of data science, Deep Neural Networks have exhibited a remarkable ability to learn complex, non-linear relationships between sets of variables. In this paper, we apply this network architecture to the task of stochastic estimation, first proposed by Adrian over 40 years ago. Two Deep Neural Networks (DNNs) are trained to estimate the pressure at selected locations in azimuthal and stream-wise arrays of pressure transducers situated just outside a Mach 0.6 jet, given instantaneous pressure measurements made at other locations in the arrays. The estimated pressure is compared with the instantaneous pressure fluctuations measured at each of the selected locations, as well as the estimates made using traditional Linear Stochastic Estimation (LSE). The root-mean-square error between the values predicted by the DNN and those measured by the transducers is shown to be nearly identical to the error associated with the traditional LSE method in the case of the azimuthal array. For the stream-wise configuration, the DNN shows an approximately 60% reduction in error over the LSE model. In addition, some limitations and possible extensions of this method are discussed.

AB - In many areas of data science, Deep Neural Networks have exhibited a remarkable ability to learn complex, non-linear relationships between sets of variables. In this paper, we apply this network architecture to the task of stochastic estimation, first proposed by Adrian over 40 years ago. Two Deep Neural Networks (DNNs) are trained to estimate the pressure at selected locations in azimuthal and stream-wise arrays of pressure transducers situated just outside a Mach 0.6 jet, given instantaneous pressure measurements made at other locations in the arrays. The estimated pressure is compared with the instantaneous pressure fluctuations measured at each of the selected locations, as well as the estimates made using traditional Linear Stochastic Estimation (LSE). The root-mean-square error between the values predicted by the DNN and those measured by the transducers is shown to be nearly identical to the error associated with the traditional LSE method in the case of the azimuthal array. For the stream-wise configuration, the DNN shows an approximately 60% reduction in error over the LSE model. In addition, some limitations and possible extensions of this method are discussed.

UR - http://www.scopus.com/inward/record.url?scp=85069005095&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85069005095&partnerID=8YFLogxK

U2 - 10.2514/6.2019-0324

DO - 10.2514/6.2019-0324

M3 - Conference contribution

AN - SCOPUS:85069005095

SN - 9781624105784

T3 - AIAA Scitech 2019 Forum

BT - AIAA Scitech 2019 Forum

PB - American Institute of Aeronautics and Astronautics Inc, AIAA

ER -