Optimale, routenbasierte Versuchsplanung (R-DoE) zur Charakterisierung technischer Systeme

Mayer, Thomas

kassel university press, ISBN: 978-3-7376-0152-8, 2016, 134 Pages

URN: urn:nbn:de:0002-401531

DOI: 10.19211/KUP9783737601535

Zugl.: Kassel, Univ., Diss. 2016

| Price and available forms -->

Content: As an enhancement of the classic DoE, this thesis proposes a route based design of experiments (R-DoE) in order to plan the entire adjustment route for automated measurements on a test rig rather than individual measurement points. The R-DoE algorithm developed within this thesis calculates an adjustment route which provides quasi-stationary measurement points for stationary data-driven simulation models, with the shortest possible total measurement time. In addition, the user's requirements, e.g. point-dependent prediction accuracies, are taken into account. Furthermore, a priori knowledge is included, e.g. technical constraints of the system to be examined. Such an adjustment route is the optimal solution of a multi-objective optimization problem and is calculated using the evolutionary algorithm (EA) implemented in the R-DoE algorithm.

The R-DoE method with the included evolutionary algorithm is illustrated by practical examples. A real validation measurement showed that the measurement time was reduced by 40% for the adjustment route optimized by the R-DoE algorithm compared with a classic DoE measurement at comparable prediction accuracy. Consequently the R-DoE method contributes to extending and further improving the efficiency of DoE methods.

Publication is available in following forms:

Full text (pdf-file, printable, Open Access - 16.10 MB)
view PDF