Identification of a Vehicle Full-scale Crash Viscoelastic System by Recursive Autoregressive Moving Average Models

This paper presents a methodology to create a recursive autoregressive model capable of estimating modal parameters of the recorded crash pulse which can be further used as parameters characterizing a physical model (spring-massdamper model). Such a viscoelastic system with nonlinear parameters estimated by RARMAX model yields results which closely follow the reference vehicle’s kinematics. This study presents the detailed description of the vehicleto-pole collision being modeled together with the derivation, establishment, and evaluation of the recursive ARMAX model. The comparative analysis of the original vehicle’s kinematics with the behavior of the Kelvin model which nonlinear parameters were estimated by the RARMAX model as well as with the behavior of the Kelvin model with constant, time-invariant parameters is presented.