Dynamic Modeling of Heavy-duty Tractor-Trailer System via Integrated Physics-Data-based Technique

In this paper, an Integrated Physics-Data-Based (IPDB) technique is applied to the nonlinear lateral dynamics modeling of a tractor-trailer system by incorporating physical principles and data. Firstly, a physical model based on Newton's law is established for the tractor-trailer system, with the force models at the hitch point and between tire-road interaction. Then, a quasi Linear-Parameter-Varying (quasi-LPV) model is converted from the nonlinear model with a carefully selected scheduling parameter vector. Finally, a data-driven LPV model is obtained by identifying the system matrices with the involving system data, which realizes the inherit integration of physics and data. The validations are conducted with simulation data extracted from Trucksim platform. The accuracy and robustness of the IPDB modeling method are investigated under various scenarios. It is demonstrated that the IPDB model can accurately capture the tractor-trailer's lateral dynamics by only using a small amount of data. For example, under double-lane change scenario, the Root Mean Square Error (RMSE) for lateral velocity and yaw rate of tractor and trailer obtained from the IPDB model are 0.0024 m/s, 0.0022 rad / s, 0.0020 m/ s, 0.0010 rad /s, respectively, compared to Trucksim sensor measurements.