三自由度四轮转向系统的神经网络模型研究

Application of Neural Network to Three Degrees of Freedom Four-wheel Steering Vehicle

车辆四轮转向的数学模型随着自由度的增加其非线性迅速增加,对于多自由度车辆模型而言,准确而迅速地求解出目标变量的值是研究的难点。针对车辆的三自由度模型,建立了带反馈的神经网络模型,通过有导师训练改变网络的权值和阈值,在输入参数后可以准确而迅速地求解出目标值。仿真结果表明,该神经网络能够很好地逼近该三自由度数学模型。

With the increasing of freedom-degree, the nonlinear of four-wheel steering model also increase sharply. As for the multi-freedom vehicle model, the most difficulty lies in getting the accurate variable goal quickly. Aimed at the three degrees of freedom vehicle model, a feedback neural network model is constructed in the research, and through the training of tutor it can change the network＇s power-value and threshold-value. After inputting parameters, it can output the goal value quickly and accurately. The result of simulation shows that it is very close to the three degrees of freedom vehicle model.