Abstract: In order to reconstruct the standard and measured vehicle road spectrum, the 1D autoregressive (AR) model of road roughness and 2D double auto-variable AR model of road roughness based on the space coordinates are built. To adopt the method of singular value decomposition of the auto-correlation coefficient matrix, the 2D double auto-variable AR model is converted into two single auto-variable models, so the complexity degree of modeling is reduced. The roughness of the class B standard road surface and the measured road surface are reconstructed on the basis of the 2D AR model, the comparison between their power spectrums verifies the correctness of the model. Based on the both AR models and Mixed Programming of Visual C++ and Matlab, the 3D virtual road reconstruction software system for vehicle is built. The frame structure of the system is given. By adding the data of road roughness to the information of the road shape, the software system can generate the 3D virtual road constituted by the 3D coordinate information of regularized nodes of road surface. After analyzing the data formats of road files of three kinds of application software, i.e. VPG, CarSim and ADAMS, the method for the system to generate road file of the said application software by using the information of 3D virtual road is presented.

Key words: 3D virtual road, AR model, Reconstruction, Road roughness, Software system