Abstract: Aiming at the problem that the steering system of the driving simulator lacks road sense feedback compared with the real vehicle, this paper designs a road sense model based on NARX neural network, and realizes the description of the friction dynamic characteristics of the steering system and the simulation of road sense. First, a driver in the loop data acquisition platform is built based on dSPACE and Carsim to realize the Controller Area Network (CAN) communication configuration of the driving simulator. Real time data of real drivers are collected, and the original data is filtered to complete the construction of NARX road sense model. Finally, the simulink/carsim joint simulation test is carried out on the road sense model. The test results show that the road sense model can effectively realize the road sense simulation function.

Key words: Driving simulator; Road sensing simulation; CAN communication; NARX neural network

摘要： 针对驾驶模拟器转向系统相对实车缺乏路感反馈这一问题，设计了一种基于 NARX 神 经网络的路感模型，实现了转向系统摩擦动态特性描述和路感模拟。首先基于 dSPACE 和 Carsim 搭建驾驶员在环数据采集平台，实现驾驶模拟器控制器局域网（CAN）通信配置。采 集真实驾驶员实时数据，并对原始数据进行滤波处理，完成 NARX 路感模型构建。最后对路 感模型进行 Simulink/Carsim 联合仿真试验。试验结果表明，该路感模型可以有效实现路感模 拟功能。

关键词: 驾驶模拟器；路感模拟；CAN 通信；NARX 神经网络