Fisher Identifiability Analysis of Longitudinal Vehicle Dynamics

Abstract

This paper examines the impact of mean-square terrain variability on longitudinal chassis parameter identifiability. This analysis is motivated by the immediate value of effective parameter estimation in various applications, including chassis model validation and active safety. Relevant literature addresses this demand through algorithms capable of estimating chassis parameters for diverse computational and on-road conditions. While the limitations of such algorithms' accuracy with respect to some driving conditions have been studied, their dependence on road grade variability remains largely unexplored. We address this open question by presenting two key contributions. First, this paper presents analytic derivations of the Fisher information matrix associated with estimating mass, drag, and rolling resistance parameters from longitudinal dynamics. We validate the analytic sensitivity expressions using simulations and experimental data gathered from an instrumented Volvo VNL300 heavy-duty freight truck. Then, this paper presents Monte Carlo simulations which illustrate the average improvements in chassis parameter identifiability associated with drive-cycles characterized by higher mean-square road grade. Our simulation studies demonstrate this result under a variety of drive cycles.