Abstract
In the automotive industry the calibration of the engine control unit is getting more and more complex because of many various boundary conditions, like the demand on CO 2 and fuel reduction. One important calibration problem is the parameter estimation for the model of the intake system of the combustion engine. This system is modeled by a physically motivated system, which can be parameterized by black-box models, like neural nets and characteristic diagrams, whose parameters must be set by an intelligent optimizer. Further, two contradictory aims must be considered and the engineer expects at the end of the optimization a pareto-front, where he can choose the best settings for the application from the pareto-optimal parameter estimations. To solve this multi-criteria optimization task an evolutionary algorithm is used, which is a combination of a genetic algorithm and an evolutionary strategy. This evolutionary algorithm is like all other stochastic searching methods leaned on the naturally biological evolution. It combines the well-known covariance matrix adaption for the mutation of the individuals with the S-metric selection for the multi-criteria fitness assignment of the individuals. It also improves these combination by the use of many subpopulations, which work parallel on various clusters, and by the use of an intelligent DoE-strategy for the initialization of the start individuals. With these improvements the developed evolutionary algorithm can easily fit the model of the intake system to test bed measurements and can provide the user a pareto-optimal set of parameters, on which he can choose on his own that ones, which he find most plausible.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Bagot, B., Pohlheim, H.: Complementary selection and variation for an efficient multiobjective optimization of complex systems. In: GECCO 2004 - Proceedings of the Genetic and Evolutionary Computation Conference (2004)
Baker, J.: Reducing Bias and Inefficiency in the Selection Algorithm. In: Proceedings of the Second International Conference on Genetic Algorithms and their Application, pp. 14–21. Lawrence Erlabaum Associates, Hillsdale (1987)
Beume, N., Naujoks, B., Emmerich, M.: SMS-EMOA: Multiobjective selection based on dominated hypervolume. European Journal of Operational Research (2006)
Christian Igel, N.H., Roth, S.: Covariance matrix adaptation for multi-objective optimization. Evolutionary Computation 15, 1–28 (2007)
Roithmeier, C.: Virtuelle Applikation von Motorsteuerungsfunktionen am Beispiel der Lasterfassunfsfunktion und der Fahrdynamikfunktionen. PhD thesis, Institut für Kolbenmaschjinen, Karlsruher Institut für Technologie (2011)
Zaglauer, S.: The Evolutionary Algorithm SAMOA with Use of Design of Experiments. In: GECCO 2012, Genetic and Evolutionary Computation Conference (2012)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2013 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Zaglauer, S., Deflorian, M. (2013). Multi-criteria Optimization for Parameter Estimation of Physical Models in Combustion Engine Calibration. In: Purshouse, R.C., Fleming, P.J., Fonseca, C.M., Greco, S., Shaw, J. (eds) Evolutionary Multi-Criterion Optimization. EMO 2013. Lecture Notes in Computer Science, vol 7811. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-37140-0_47
Download citation
DOI: https://doi.org/10.1007/978-3-642-37140-0_47
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-37139-4
Online ISBN: 978-3-642-37140-0
eBook Packages: Computer ScienceComputer Science (R0)