Design of EGR Solenoid Valve Controller using Neural Networks
Design of EGR Solenoid Valve Controller using Neural Networks
This paper describes an initial investigation into the use of neural network learning algorithms to obtain a controller for a non-linear system over a large operating space within the context of automotive applications. In order to perform a comparative study of the various adaptive systems, the problem of controlling the motion of a solenoid-operated EGR (Exhaust Gas Recirculation) valve is considered. This study also compares a neurocontroller with a PID controller for various position step changes in both directions. During the investigation it was found that the performance of the neurocontroller was consistently better, particularly for large demanded step changes, and that the neurocontroller consistently used less control energy. Further work will focus on why these nonlinear learning systems outperform perform PID controllers in this application.
985--989
Arain, M.A.
464b59b9-ac68-44d0-9119-d07033e28c53
Brown, M.
52cf4f52-6839-4658-8cc5-ec51da626049
1995
Arain, M.A.
464b59b9-ac68-44d0-9119-d07033e28c53
Brown, M.
52cf4f52-6839-4658-8cc5-ec51da626049
Arain, M.A. and Brown, M.
(1995)
Design of EGR Solenoid Valve Controller using Neural Networks.
EUFIT '95.
.
Record type:
Conference or Workshop Item
(Other)
Abstract
This paper describes an initial investigation into the use of neural network learning algorithms to obtain a controller for a non-linear system over a large operating space within the context of automotive applications. In order to perform a comparative study of the various adaptive systems, the problem of controlling the motion of a solenoid-operated EGR (Exhaust Gas Recirculation) valve is considered. This study also compares a neurocontroller with a PID controller for various position step changes in both directions. During the investigation it was found that the performance of the neurocontroller was consistently better, particularly for large demanded step changes, and that the neurocontroller consistently used less control energy. Further work will focus on why these nonlinear learning systems outperform perform PID controllers in this application.
This record has no associated files available for download.
More information
Published date: 1995
Additional Information:
Organisation: Aachen, Germany
Venue - Dates:
EUFIT '95, 1995-01-01
Organisations:
Electronics & Computer Science
Identifiers
Local EPrints ID: 250149
URI: http://eprints.soton.ac.uk/id/eprint/250149
PURE UUID: bd08cbcf-da2d-4ce3-a956-d7b3e09e8d20
Catalogue record
Date deposited: 04 May 1999
Last modified: 10 Dec 2021 20:07
Export record
Contributors
Author:
M.A. Arain
Author:
M. Brown
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics