Research Papers: Gas Turbines: Combustion, Fuels, and Emissions

Accelerometer-Based Combustion Metrics Reconstruction With Radial Basis Function Neural Network for a 9 L Diesel Engine

[+] Author and Article Information
Libin Jia

e-mail: libinj@mtu.edu

Jeffrey Naber

e-mail: jnaber@mtu.edu

Jason Blough

e-mail: jrblough@mtu.edu
Department of Mechanical Engineering-
Engineering Mechanics,
Michigan Technological University,
1400 Townsend Dr.,
Houghton, MI 49931

Seyed Alireza Zekavat

Department of Electrical and
Computer Engineering,
Michigan Technological University,
1400 Townsend Dr.,
Houghton, MI 49931
e-mail: rezaz@mtu.edu

Contributed by the Combustion and Fuels Committee of ASME for publication in the JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER. Manuscript received July 3, 2013; final manuscript received October 18, 2013; published online November 19, 2013. Assoc. Editor: Song-Charng Kong.

J. Eng. Gas Turbines Power 136(3), 031507 (Nov 19, 2013) (9 pages) Paper No: GTP-13-1231; doi: 10.1115/1.4025886 History: Received July 03, 2013; Revised October 18, 2013

Accelerometer-based combustion sensing in diesel engines has the potential of providing feedback for combustion control to reduce fuel consumption and engine emissions at a lower cost than in-cylinder pressure sensors. In this work, triaxial block-mounted accelerometers were used to measure the engine vibration, and pressure transducers were installed to measure the in-cylinder pressure. The in-cylinder pressure can be further utilized to compute combustion metrics, including the apparent heat release rate (AHR). Engine tests were conducted for various speeds, torques, and start of injections, on a 9 L in-line six-cylinder diesel engine equipped with a common rail high pressure injection system. The relationship between engine block acceleration and AHR was modeled using a radial basis function neural network (RBFNN). By inputting the accelerometer signal to the fixed network, AHR and other combustion metrics were estimated. As the primary concern for radial basis network training is the hidden layer weight vector selection, two algorithms for weight vector selection (modified Gram–Schmidt orthogonalization and principal component analysis) were evaluated by examining the robustness of the resulting network. One-third of the conducted tests were utilized to train the network. The network was then applied to estimate the AHR for the remaining validation tests which were not used to train the network. Comparisons were made based on the combustion metrics estimation results and the selection efficiency among the two weight vector selection methods and the random selection method. Moreover, the capability concerning the network's tolerance for additive noise was also investigated. Results confirmed that the modified Gram–Schmidt method achieved much more accurately estimated combustion metrics with the highest efficiency. On the basis of this study, a real-time closed-loop control strategy was proposed with the feedback provided based on the application of the trained RBFNN.

Copyright © 2014 by ASME
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Fig. 1

Sensor placement on the 9 L diesel engine

Grahic Jump Location
Fig. 2

Test conditions (87 test points) comprised of load conditions

Grahic Jump Location
Fig. 3

Test conditions including breakdown of training and validation data plotted as start of injection versus speed (a) and torque (b)

Grahic Jump Location
Fig. 4

AHR trace marked with combustion metrics (engine speed = 2200 rpm, engine load = 725 Nm, SOI = −11.3 deg CA)

Grahic Jump Location
Fig. 5

Radial basis function network structure

Grahic Jump Location
Fig. 6

Accelerometer signal overlaid upon derived AHR trace

Grahic Jump Location
Fig. 7

Schematic illustration for selection of Xk from Q

Grahic Jump Location
Fig. 8

AHR signatures indicating PACL differences between estimated and derived for engine speed = 2200 rpm and engine load = 725 Nm

Grahic Jump Location
Fig. 9

Comparison of AHR estimation results based on three RW selection methods

Grahic Jump Location
Fig. 10

Frequency content for accelerometer signal and artificial noise

Grahic Jump Location
Fig. 11

Artificial noise in time domain

Grahic Jump Location
Fig. 12

Computation time comparison between PCA and MGS methods

Grahic Jump Location
Fig. 13

Start of injection closed-loop control with CA50 as the feedback




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