Introducing Functional Data Analysis to Coast Down Modeling for Rolling Resistance Estimation 2015-01-9111
Coast-down modeling has been widely used to assess vehicle aerodynamic drag and rolling resistance by fitting a vehicle resistance model to speed measurements and thereby get an estimate on model parameters. Here a coast-down model is used for assessing how road surface characteristics influence rolling resistance. Parameter estimation as well as an extensive perturbation analysis of the parameter fit with respect to data noise has been performed. Functional Data Analysis (FDA) is introduced and discussed as a tool for this. It is concluded that FDA is a powerful tool for 1) approximating derivatives, 2) assessing the degree of smoothing of the data 3) handling noise sources in the perturbation analysis and 4) enabled numerical solutions of the coast-down Ordinary Differential Equation (ODE) model. Investigations showed that MPD was the most important parameter compared to IRI although MPD data required smoothing for optimal model fit. Furthermore, it is concluded that the model responds nicely to the statistical tests in the perturbation analysis. However, certain parameters associated with surface related rolling resistance were unstable in the sensitivity tests.
Citation: Andersen, L. and Larsen, J., "Introducing Functional Data Analysis to Coast Down Modeling for Rolling Resistance Estimation," SAE Int. J. Passeng. Cars - Mech. Syst. 8(2):786-796, 2015, https://doi.org/10.4271/2015-01-9111. Download Citation
Author(s):
Lasse G. Andersen, Jesper K. Larsen
Affiliated:
Roskilde University Center
Pages: 11
ISSN:
1946-3995
e-ISSN:
1946-4002
Also in:
SAE International Journal of Passenger Cars - Mechanical Systems-V124-6EJ, SAE International Journal of Passenger Cars - Mechanical Systems-V124-6, SAE International Journal of Passenger Cars - Mechanical Systems-V125-6
Related Topics:
Drag
Simulation and modeling
Fittings
Statistical analysis
SAE MOBILUS
Subscribers can view annotate, and download all of SAE's content.
Learn More »