Due to complex form and difficult parameters fitting of the traditional tire mechanics model, it is not convenient for users to design vehicle dynamic control system. In this work, a data-driven modelling method for the tire longitudinal-slip mechanical characteristics is proposed based on piecewise affine identification approach. Firstly, the experimental data, which can accurately reflect the tire nonlinear longitudinal-slip mechanical characteristics under the conditions, where the tire vertical loads are respectively 3 124, 6 530, 8 036, 9 468 and 11 760 N, the tire pressure is 880 kPa, the variation range of tire longitudinal slip ratio is -1-0.5, and the road peak adhesion coefficients are 0.34 and 0.77, are obtained through a flat-plate test bench. On this basis, by analyzing the experimental data, the general mapping relationship between the tire longitudinal force and its main influencing factors is determined; and then effective identification of the tire longitudinal-slip mechanical characteristics can be conducted by adopting the piecewise affine identification theory. The aforementioned identification procedure mainly consists of experimental data clustering, parameters estimation of affine submodels, and computation of the hyperplane coefficient matrices, which are accomplished by using the algorithms such as modified K-means, least square, and fuzzy weighted proximal support vector machines, respectively. Finally, 10 affine submodels are obtained for low adhesion road and high adhesion road, respectively, to fit the experimental data of the tire nonlinear longitudinal-slip mechanical characteristics. To verify the accuracy of the identified tire model, the model simulation data are compared with the experimental data. The comparison results show that the overall fitting accuracy of the model is more than 95%, thus the identified model can approximate the tire nonlinear longitudinal-slip mechanical characteristics with high accuracy. © 2021, Editorial Office of Journal of Xi'an Jiaotong University. All right reserved.