CAJ | 학술논문

为提高智能车辆自主循迹控制的精度，提出了一种基于模糊神经网络控制( FNNC)和神经网络预测( NNP)的智能循迹控制策略。转向控制器的输入量有3个：预瞄点处的横向循迹误差、汽车横摆角速度和侧向加速度。车速控制器的输入量有4个：预瞄点处的面积误差、侧向加速度、汽车侧偏角和转向盘转角。网络训练采用误差反向传播法。仿真与试验结果表明，所设计的循迹控制器通过对驾驶员操作样本的训练，能实现对车辆的车速与转向控制，横向循迹误差和目标车速均比较理想。
위제고지능차량자주순적공제적정도，제출료일충기우모호신경망락공제( FNNC)화신경망락예측( NNP)적지능순적공제책략。전향공제기적수입량유3개：예묘점처적횡향순적오차、기차횡파각속도화측향가속도。차속공제기적수입량유4개：예묘점처적면적오차、측향가속도、기차측편각화전향반전각。망락훈련채용오차반향전파법。방진여시험결과표명，소설계적순적공제기통과대가사원조작양본적훈련，능실현대차량적차속여전향공제，횡향순적오차화목표차속균비교이상。
[ Abstract] In order to improve the autonomous path tracking control accuracy of intelligent vehicle, an in-telligent path tracking control strategy is proposed based on fuzzy neural network control and neural network predic-tion. The inputs of steering controller are the transverse path tracking error at preview points and the yaw rate and lateral acceleration of vehicle, while those for speed controller are the area error at preview points and the lateral ac-celeration, side slip angle and steering wheel angle of vehicle, and error back propagation technique is adopted for network training. The results of simulation and test show that through the training of driver operation samples, the path tracking controller designed can realize speed and steering control of intelligent vehicle with relatively desirable transverse path tracking error and target speed.