Simulation Study of the Impact of Road Terrain Variability on Identifiability of Longitudinal Vehicle Chassis Parameters

Open Access
Author:
Kandel, Aaron Isaac
Area of Honors:
Mechanical Engineering
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
  • Hosam K. Fathy, Thesis Supervisor
  • Daniel Humberto Cortes Correales, Honors Advisor
Keywords:
  • parameter identifiability
  • parameter estimation
  • system identification
  • system dynamics
Abstract:
This thesis applies a basic simulation study to demonstrate the relationship between mean-square road terrain variability and vehicle chassis parameter identifiability. This work is motivated by the importance of chassis parameter estimation in proper system identification, specifically parameterizing vehicle dynamics models with unknown system parameters and designing vehicle road tests which are optimized for chassis parameter identifiability. The study of vehicle chassis parameter estimation spans decades in the literature, and has been performed with the use of varying chassis dynamics models and estimation algorithms including simple linear least-squares regression. However, the quantification of the impact of terrain variability on chassis parameter identifiability remains largely unexplored. This work illustrates this relationship by performing simple linear least-squares estimation on a series of simulated driving cycles which exhibit progressively scaled sinusoidal terrain profiles. The Cramer-Rao lower bounds for the errors of the resulting parameter estimates are obtained through simple Fisher information analysis. A terrain variability metric is defined based on the mean-square of road grade across each driving cycle, and this metric is related to the estimation error values in a series of plots. The results from this simulation study demonstrate the monotonic and decreasing relationship between terrain variability and chassis parameter identifiability for a sinusoidal terrain profile. The significant degree to which terrain variability is shown to increase the quality of chassis parameter estimates can motivate the design of future experiments for estimating vehicle chassis parameters using data from on-road experiments.