Abstract
Handling and cornering behaviour are both important aspects of lateral dynamics response of modern cars: nowadays handling performances of vehicles are investigated during time-expensive road test sessions, where cars are tested in transient and steady-state response to standard manoeuvres. Numerical analysis of vehicle lateral dynamics may represent a valid shortcut to experimental sessions: road tests can be reduced and the data collected used to identify and validate a suitable vehicle model for estimating the response to different steering-wheel inputs. The capability of mathematical models to accurately simulate the real behaviour of a car depends largely on the knowledge of all the coefficients involved in the complete description of the model itself, in particular those related to the formulation of tyre cornering forces vs. slip angle characteristic. The paper shows a method for the identification of unknown coefficients set that appears in a simple single-track dynamic vehicle model, using Pacejkas Magic Formula with tyre relaxation length; the purpose of this work is to develop an original technique based on an Extended Kalman Filter (EKF) applied to non-linear system identification, where non-linearity is due to tyre cornering forces. Data collected during several road tests (e.g. step-steering input at different levels) are provided simultaneously to EKF, in order to correctly estimate non-linear tyre cornering forces characteristic in the whole range of possible slip angle variation. The technique has been completely validated first by estimating unknown coefficients set starting from a small number of step-steering input, then simulating different kinds of tests (such as lane change and steering-pad) with that identified set and matching the results of simulations with the corresponding data collected on road.