Identification of the tire model and road adhesion coefficient based on vehicle dynamic response data and the BP neural network algorithm

The longitudinal and lateral dynamic responses of a vehicle are essentially determined by the forces resulting from the contact between the tire and road surface. Therefore, identifying the tire model and road adhesion coefficient plays an important role in vehicle dynamics control. At present, the identification of a nonlinear tire model essentially assumes that slip ratio-longitudinal force data or wheel side slip angle-lateral force data are known and that nonlinear fitting is performed, and the road adhesion coefficient based on the dynamic response is also identified on the premise of a known tire model. In this paper, the characteristic of approximating arbitrary nonlinear mapping relationships by a BP neural network is taken advantage of, and an off-line identification method of the Dugoff tire model based on onboard sensors and vehicle dynamics response is proposed. On this basis, an online identification method for the road adhesion coefficient is subsequently proposed. Finally, vehicle experiments on steering and braking are performed to verify the accuracy of the method in the Simulink-CarSim joint simulation environment.