基于神经网络方法的台试轮胎滚动阻力模型研究

Research on model of bench-based tire rolling resistance with neural network method

采用反拖测试方法,在双滚筒式底盘测功机上进行了大量有关轮胎滚动阻力影响因素(包括胎压、速度、载荷、胎温及轮胎的型号等)的交叉试验。以这些因素作为神经网络的训练输入参数,建立了台试轮胎滚动阻力神经网络模型。通过试验验证,用神经网络模型预估基于双滚筒测试的轮胎滚动阻力误差小于4%,为利用底盘测功机准确测试汽车动力性等工作奠定了基础。

In the paper, a lot of experimentations about affecting factors of tire rolling resistance have been done on dualroller chassis dynamometer with contrarydriving test method. These factors include tire pressure, speed, load, tire temperature and different type tires.Based on the testing data, the neural network model of rolling resistance is set up in the paper. The experiment result shows that the tire rolling resistance error from the neural network model is less than 4%. Authors establish the foundation for automobile dynamic property testing.