Click on “Download PDF” for the PDF version or on the title for the HTML version.


If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Methodology to Perform Identifiability Analysis for Off-Road Vehicle Tire-Soil Parameter Estimation

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  2011 Louisville, Kentucky, August 7-10, 2011  1111526.(doi:10.13031/2013.38126)
Authors:   Simon L Nielsen, Manoj Karkee, Brian L Steward
Keywords:   Tractor and implement model, parameter uncertainty, parameter identifiability, parameter identification, profile likelihood, model-based design

The growing trend of model-based design in off-road vehicle engineering requires models that are sufficiently accurate if they are to be used with confidence. Uncertain model parameters are often identified from measured data by using an optimization procedure, but it is important to understand the limitations of such a procedure and to have methods available for assessing the uniqueness and confidence of the results. Model identifiability analysis is used to determine whether system measurements contain enough information to estimate the model parameters. A numerical approach based on the profile likelihood of parameters was utilized to evaluate the local structural and practical identifiability of a tractor and single axle towed implement model with six uncertain tire force model parameters from tractor and implement yaw rate data. The analysis considered simulated data with known model parameter values to examine the effect of measurement error on the identifiability. The accuracy and confidence of identification tended to decrease as the quality of the data decreased, to the point that five of the six parameters were considered practically unidentifiable from the information available. Overall, the study showed that experimental factors such as noise can affect the amount of information available in a dataset for identification and that error in the measured data can propagate to error in model parameter estimates.

(Download PDF)    (Export to EndNotes)