Experimental Identification of Road-Vehicle Dynamics Using Autoregression

This paper presents an identification technique, for the road - vehicle dynamic behavior of suspension systems, by implementing an autoregressive system with exogenous input (ARX). The ARX model was proposed as a simple and powerful tool, in terms of accuracy and computational time, compared to the complexity and significant computational cost involved with the neural networks approach which is commonly used. An experimental approach is introduced based on training data being extracted from sensors readings which are attached to specific locations, of a real car suspension, in an attempt to capture the dynamic behavior of a quarter car model. In addition, two different ARX models were created, once by using front-left wheel excitation only and another by front and rear wheels excitations. It is found that the ARX model, based on measurements extracted from only one wheel of a real car suspension, could accurately represent the vertical dynamics of the whole vehicle.