0
Research Papers

Diesel Engine Acoustic Emission Airflow Clogging Diagnostics With Machine Learning

[+] Author and Article Information
Jim Cowart

U.S. Naval Academy,
Annapolis, MD 21402
e-mail: cowart@usna.edu

Patrick Moore

U.S. Naval Academy,
Annapolis, MD 21402
e-mail: m194494@usna.edu

Harrison Yosten

U.S. Naval Academy,
Annapolis, MD 21402
e-mail: m18712@usna.edu

Leonard Hamilton

U.S. Naval Academy,
Annapolis, MD 21402
e-mail: ljhamilt@usna.edu

Dianne Luning Prak

U.S. Naval Academy,
Annapolis, MD 21402
e-mail: prak@usna.edu

Manuscript received March 12, 2019; final manuscript received March 27, 2019; published online April 15, 2019. Editor: Jerzy T. Sawicki.This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. Approved for public release; distribution is unlimited.

J. Eng. Gas Turbines Power 141(7), 071021 (Apr 15, 2019) (9 pages) Paper No: GTP-19-1119; doi: 10.1115/1.4043332 History: Received March 12, 2019; Revised March 27, 2019

A diesel engine electrical generator set (“gen-set”) was instrumented with in-cylinder indicating sensors as well as acoustic emission microphones near the engine. Air filter clogging was emulated by progressive restriction of the engine's inlet air flow path during which comprehensive engine and acoustic data were collected. Fast Fourier transforms (FFTs) were analyzed on the acoustic data. Dominant FFT peaks were then applied to supervised machine learning neural network analysis with matlab-based tools. The progressive detection of the air path clogging was audibly determined with correlation coefficients greater than 95% on test data sets for various FFT minimum intensity thresholds. Further, unsupervised machine learning self-organizing maps (SOMs) were produced during normal-baseline operation of the engine. The degrading air flow engine sound data were then applied to the normal-baseline operation SOM. The quantization error (QE) of the degraded engine data showed clear statistical differentiation from the normal operation data map. This unsupervised SOM-based approach does not know the engine degradation behavior in advance, yet shows clear promise as a method to monitor and detect changing engine operation. Companion in-cylinder combustion data additionally shows the degrading nature of the engine's combustion with progressive airflow restriction (richer and lower density combustion).

Copyright © 2019 by ASME
Your Session has timed out. Please sign back in to continue.

References

Wong, K. I. , Wong, P. K. , Cheung, C. S. , and Vong, C. M. , 2013, “ Modeling and Optimization of Biodiesel Engine Performance Using Advanced Machine Learning Methods,” Energy, 55(15), pp. 519–528. [CrossRef]
Albarbar, A. , Gu, F. , and Ball, A. D. , 2010, “ Diesel Engine Fuel Injection Monitoring Using Acoustic Measurements and Independent Component Analysis,” Measurement, 43(10), pp. 1376–1386. [CrossRef]
Sharkey, A. J. C. , Chandroth, G. O. , and Sharkey, N. E. , 2000, “ A Multi-Net System for the Fault Diagnosis of a Diesel Engine,” Neural Comput. Appl., 9(2), pp. 152–160. [CrossRef]
Fischer, M. , Boettcher, J. , Kirkham, C. , and Georgi, R. , 2009, “ OBD of Diesel EGR Using Artificial Neural Networks,” SAE Paper No. 2009-01-1427.
Ghaboussi, J. , and Banan, M. R. , 1994, “ Neural Networks in Engineering Diagnostics,” SAE Paper No. 941116.
Huang, R. , Xi, L. , Li, X. , Richard Liu, C. , Qiu, H. , and Lee, J. , 2007, “ Residual Life Predictions for Ball Bearings Based on Self-Organizing Map and Back Propagation Neural Network Methods,” Mech. Syst. Signal Process., 21(1), pp. 193–207. [CrossRef]
Kohonen, T. , Oja, E. , Simula, O. , Visa, A. , and Kangas, J. , 1996, “ Engineering Applications of the Self-Organizing Map,” Proc. IEEE, 84(10), pp. 1358–1384. [CrossRef]
Alhoniemi, E. , Hollmen, J. , Simula, O. , and Vesanto, J. , 1999, “ Process Monitoring and Modeling Using the Self-Organizing Map,” Integr. Comput.-Aided Eng., 6(1), pp. 3–14. [CrossRef]
Ypma, A. , and Duin, R. P. W. , 1997, “ Novelty Detection Using Self-Organizing Maps,”Progress in Connectionist-Based Information Systems, Vol. 2, Springer, New York, pp. 1322–1325.
Ball, A. D. , Gu, F. , and Li, W. , 2000, “ The Condition Monitoring of Diesel Engines Using Acoustic Measurements—Part 2: Fault Detection and Diagnosis,” SAE Paper No. 2000-01-0368.
Luning Prak, D. , Luning Prak, P. J. , Trulove, P. , and Cowart, J. S. , 2016, “ Formulation of Surrogate Mixtures Based on Physical and Chemical Analysis of Hydrodepolymerized Cellulosic Diesel Fuel,” Energy Fuels, 30(9), pp. 7331–7341. [CrossRef]
Chun, K. M. , and Heywood, J. B. , 1987, “ Estimating Heat Release and Mass of Mixture Burned From SI Engine Pressure Data,” Combust. Sci. Technol., 54(1–6), pp. 133–143. [CrossRef]
Gatowski, J. A. , Balles, E. N. , Chun, K. M. , Nelson, F. E. , Ekchian, J. A. , and Heywood, J. B. , 1984, “ Heat Release Analysis of Engine Pressure Data,” SAE Paper No. 841359.
MathWorks, Nonlinear Regression,” MathWorks, Natick, MA, accessed Mar. 26, 2019, https://www.mathworks.com/discovery/nonlinear-regression.html
MathWorks, “Fit Data With a Shallow Neural Network,” MathWorks, Natick, MA, accessed Mar. 26, 2019, https://www.mathworks.com/help/nnet/gs/fit-data-with-a-neural-network.html
MathWorks, “Improve Shallow Neural Network Generalization and Avoid Overfitting,” MathWorks, Natick, MA, accessed Mar. 26, 2019, https://www.mathworks.com/help/nnet/ug/improve-neural-network-generalization-and-avoid-overfitting.html
Kohonen, T. , 1995, Self-Organizing Maps (Springer Series in Information Sciences), Vol. 30, Springer, Berlin.
Laboratory of Computer and Information Science — Adaptive Informatics Research Centre, 2019, “Homepage of SOM Toolbox,” Helsinki University of Technology, Espoo, Finland, accessed Mar. 26, 2019, http://www.cis.hut.fi/somtoolbox/
Laboratory of Computer and Information Science — Adaptive Informatics Research Centre, 2019, “Documentation,” Helsinki University of Technology, Espoo, Finland, accessed March 26, 2019, http://www.cis.hut.fi/somtoolbox/documentation/

Figures

Grahic Jump Location
Fig. 1

Air filter effective flow coefficient

Grahic Jump Location
Fig. 2

Operating engine equivalence ratio data as a function of air inlet clogging percentage

Grahic Jump Location
Fig. 3

Engine heat release characteristics as a function of inlet air clogging percentage

Grahic Jump Location
Fig. 4

In-cylinder measured pressure and heat release rate for baseline, 4 of 9 (4/9) and 8 of 9 (8/9) clogged inlet

Grahic Jump Location
Fig. 5

Full data set FFT at various inlet blockage levels for the data collected with the GRAS microphone

Grahic Jump Location
Fig. 6

Full data set FFT at various inlet blockage levels for the data collected with the laptop microphone

Grahic Jump Location
Fig. 7

Full data set FFT (minute in continuous data) compared against the 60 individual discrete FFT results for the data from the GRAS microphone

Grahic Jump Location
Fig. 8

Full data set FFT (minute is continuous data) compared against the 60 individual discrete FFT results for the data from the laptop microphone

Grahic Jump Location
Fig. 9

Sample supervised neural network schematic

Grahic Jump Location
Fig. 10

Training results from laptop sound data with 20% FFT threshold and 10 hidden nodes in neural network

Grahic Jump Location
Fig. 11

Test results from laptop sound data with 20% FFT threshold and ten hidden nodes in neural network

Grahic Jump Location
Fig. 12

Correlation coefficient (R) results (top) from both microphones at various FFT threshold levels, and RMSE bottom

Grahic Jump Location
Fig. 13

SOM unsupervised neural network schematic

Grahic Jump Location
Fig. 14

Mapping of SOM nodes with “B” placed in nodes (left) were baseline data has a BMU. Baseline SOM (middle) and clogging data included (right).

Grahic Jump Location
Fig. 15

SOM BMU vector versus actual data vector for baseline operation

Grahic Jump Location
Fig. 16

SOM BMU vector versus actual data vector for 8/9 restricted operation

Grahic Jump Location
Fig. 17

Mean quantization errors for the laptop and GRAS data analyzed with a SOM and 90% CIs

Grahic Jump Location
Fig. 18

SOM created from all five data sets with various airflow clogging

Tables

Errata

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In