TU Berlin

Chair of Electronic Measurement and
Diagnostic Technology

Page Content

to Navigation

Publications from the Chair of Electronic Measurement and Diagnostic Technology

Data Based Modelling of Engine Dynamics
Citation key DOEKONFERENZ2007
Author Nahrstaedt, H and Gühmann, C
Title of Book Design of Experiments (DOE) in Engine Development, III
Year 2007
Publisher expert Verlag
Edition erste
Abstract The rapid calibration of automotive electronic control units (ECU), the on-board diagnosis and the control of the power train and its components require more and more sophisticated identification methods for highly non-linear dynamic systems. Besides the good model accuracy the models has to run on microprocessors or PCs in real-time applications. We describe in this paper an identification method based on a Wiener model structure for non-linear multi-input single-output systems. The used Wiener models consist of a dynamic part, represented by linear transfer functions and the nonlineari-ties, which are modelled with polynomials or neural networks. The model structure, the parameter optimisation with the Gauss-Newton method, tests of goodness of fit as well as the strengths and drawbacks are shown. As a practical example the identification of the temperature in dependence of five inputs (advance angle, engine speed, load, input valve angle and output valve angle) is discussed.
Download Bibtex entry


Quick Access

Schnellnavigation zur Seite über Nummerneingabe