汽车四轮转向自适应模糊神经网络控制研究

为了实现现实车辆运动的多自由度和非线性，在Simulink环境下建立包含车辆侧倾运动和轮胎非线性的三自由度四轮转向模型，针对大多控制方法需要依赖被控对象为精确数学模型的缺陷，提出具有联想、自学习、自识别、自适应特性的自适应模糊神经网络四轮转向控制策略；通过以前轮转角及车速作为输入，并依此确定后轮转角的输出，建立获得训练样本的仿真实验模型，用混合法训练得到自适应模糊神经网络控制器，并分别与前轮转向、比例控制和横摆角速度反馈控制下的四轮转向控制器进行仿真比较分析．结果表明自适应模糊神经网络控制使车辆在低速到中、高速时质心侧偏角趋于零，具有较强的鲁棒性；在角阶跃、移线实验中，控制效果优于前轮转向、比例控制和横摆角速度反馈控制，较大地改善了车辆的操纵性能．

Research on adaptive fuzzy neural network control of four-wheel steering system

In order to achieve the multi degree of freedom and nonlinear movement of the vehicle， three degrees of freedom and four wheel steering model was built up under Simulink environment. For most control methods need to rely on the defect that the controlled object should be an accurate mathematical model， we put up the fuzzy neural network which is self associated， self learning， self recognition and adaptive for four wheel steering control strategy. With rotor angle of front wheel and speed of the vehicle as input， the rotor angle of rear wheel as the output， we built the simulation model for the training. Hybrid method was used to acquire the sample self adaptive fuzzy neural network controller， and compared the controller with that which is obtained under the feedback control of the front wheel steering， proportional control and yawing angular velocity. Results show that the self adaptive fuzzy neural network control makes the vehicle mass center side slip angle tend to zero when the speed is from low to high， and it has strong robustness. In angle step， line movement experiments， the control effect is better than that of the front wheel steering and proportional control and yawing angular velocity feedback control， it greatly improves the control performance of the vehicle．