Online Vehicle Parameter Identification Using Wide-Array of Nonlinear Dynamics Approximation

One of the main challenges in vehicle modeling is accurately determining the parameters of the systems. To address this issue, most methods rely on variants of Recursive Least Squares (RLS), Kalman filter, and particle filter algorithms. This study investigates the implementation of a new method called Wide-Array of Nonlinear Dynamics Approximation (WyNDA) to estimate five parameter values of a vehicle in real-time: two parameters determining the longitudinal position of the Center of Gravity (CoG), two stiffness coefficients between the front and rear wheels and the road, and the yaw moment of inertia. We conducted two simulation cases with constant and varying vehicle speeds and analyzed the algorithm's effectiveness numerically. The simulation results revealed that each parameter converges to its actual value, resulting in substantially minor errors in the vehicle state estimation.