Online Optimization
Siemens VDO, Regensburg
From 01/2000 - 02/2002: Calibration Methods and Tools (partly 50% with other offline activity)
03/2002 to 01/2008 : Test Center, Engine Test Bench
Technical Project Leader for Online Optimization and Online Design of Experiments
Motivation:
"Manufacturers of modern internal combustion engines are confronted with increasing challenges: The legal restrictions concerning exhaust emissions are getting stricter, fuel is becoming more expensive and customers are demanding powerful and comfortable engines with low fuel consumption. As a result, engine complexity has greatly increased, as has the number of adjustable parameters. Tuning the engine thus results in a complex optimisation problem, whose solution requires the aid of computers. The application of computer-aided offline optimisation has become common. (...) They rely on the following principle: On the basis of an experimental plan, a sufficiently large number of measurements are performed at the test bed. This data, possibly combined with previous knowledge, is used to calculate a computer model that simulates the engine. This model can then be optimised by appropriate algorithms. The optima are again verified at the test bed and finally used for tuning the engine. In the online optimisation, the optimisation system directly interacts with the test bed, which means that, in the ideal case, the process is fully automatic. This approach carries the further advantage that information gained by measurement can be evaluated immediately. (...) Offline optimisation, on the other hand, has the advantage that the computation time can be completely separated from the test bed, i.e. the test bed can be used manually during the computations. The automatic test bed control with appropriate limit handling poses new challenges to the algorithms."
From MTZ 2003/05, BMW, Model-based Online-Optimization of Modern Internal Combustion Engines.
Development of a Software for running Online Optimization and Design of Experiments at the Engine Test Bench.
Platform interfacing MATLAB, Morphée and INCA.
(Optimization algorithm description available in ENBIS paper).
Publications / Conferences:
ENBIS (European Netwrok for Business and Industrial Statistics ) Conference 2007: Abstract
Workshop on Parameter Estimation and Optimal Design of Experiments - PARAOPE in Heidelberg in 2004 entitled: "The Use of intelligent Experimental Designs for Optimal Automotive Engine Calibration Online at Engine Test Bench"
Optimisation en ligne de moteurs automobiles sur bancs d'essais. Revue Techniques Avancées, June 2004. French
Second Workshop on Non-linear Optimization “Theoretical Aspects of Surrogate Optimization”, Coimbra 2002-05-16 I made a presentation entitled: “The Use of Surrogate-based Optimization in Engine Automatic Calibration on Test Bench”.
See https://www.gerad.ca/Charles.Audet/PUB/OPTE.doc.
http://studylib.net/doc/7449127/2-the-engine-calibration-optimization-problem OPTE Special Issue on Surrogate Optimization
Bibliography:
Full bibliography (.bib file, last update 2007)
IAV DOE Conference
BMW MTZ Paper 05/2003
Model-Based Online Optimisation of Modern Internal Combustion Engines
Part 1: Active Learning (Deutsch)
Part 2: Limits of the Feasible Search Space (Deutsch)
Modellbasierte Online-Optimierung in der Simulation und am Motorenprüfstand MTZ 01/2007
Mark Abramson PhD Thesis, Pattern Search Algorithms for Mixed Variable General Constrained Optimization Problems. (2002)
Related Software:
NOMADm Optimization Software (see Acknowledgements) Mark Abramson
Acknowledgements
I owe a special debt of gratitude to my colleagues, John Dennis and Charles Audet, whose assistance in this project has been invaluable. I would also like to thank Keith Berrier, Olga Brezhneva, Ana Custodio, Gilles Couture, Thierry Dalon, John Dunlap, Nikolaus Hansen, Arantzazu Garcia-Lekue, Alison Marsden, Rachael (Pingel) Robison, Jacob Sondergaard, and Luis Vicente for numerous suggestions for correcting and improving the code. Also, thanks to Rakesh Kumar at MathWorks, Inc. for helpful discussions, and thanks to several other users, who have identified bugs and patiently waited for the fix.
DACE: http://www2.imm.dtu.dk/projects/dace/ Jacob Søndergaard, Hans Bruun Nielsen
https://github.com/l2/Biblio-code/blob/master/meta/dace/changelog
Version 1.3, July 3, 2002
c We would like to acknowledge the comments and suggestions from Thierry Dalon, from Siemens VDO Automotive AG, which is a great help in improving this software.
Commercial Software:
Engine Test Bench Automation Software:
MathWorks Genetic Algorithm and Direct Search Toolbox
MathWorks Model-Based Calibration Toolbox
Keywords:
DOE: Design of Experiment
Model-based Calibration
Surrogate-based optimization
GPS: Generalized Pattern Search
FrontPage
Back to Engine Test Bench
Version 1.3, July 3, 2002 |
|
|
|
* A developer debug statement in PREDICTOR line 52: |
|
|
|
save NX x |
|
|
|
is removed. |
|
|
|
+ DSMERGE is a new function for merging data sets with multiple |
|
design sites (i.e., sites sampled more than one time). |
|
See 'help dsmerge' for more information. |
|
|
|
+ PREDICTOR now return Jacobian matrices for model and mean |
|
squared error also for cases with multiple response functions. |
|
|
|
c We would like to acknowledge the comments and suggestions from |
|
Thierry Dalon, from Siemens VDO Automotive AG, which is a great |
|
help in improving this software. |
Comments (0)
You don't have permission to comment on this page.