汽车转向操纵的单神经元自适应控制_高振海.pdfVIP

34 3 吉林 大学 学 报 ( 工学 版) Vol. 34 No. 3 2004 7 Jo rnal of Jilin University (Engineering and Technology Edition) J ly 2004 : 1671- 5497(2004) 03- 0362- 05 1 2 高振海 , 李向瑜 ( 1. 吉林大学 汽车动态模拟国家重点实验室, 吉林 长春 130022; 2. 吉林大学 汽车工程学院, 吉林 长春 130022) : 基于汽车操纵稳定性总方 评价方法和神经网络控制理论, 建立了汽车转向单神经元 自适应PID 控制算法利用操纵稳定性评价中的轨迹跟随误 和方向盘忙碌程度的评价指 标, 建立了神经元学习的二次型性能指标函数, 并采用梯度下降算法实现了单神经元控制器的 连接权值的调整最终采用汽车7 自由度非线性动力学模型进行了不同行驶速度下的仿真计 算结果表明: 该控制算法可较为精确地控制汽车对预期行驶轨迹的转向运动, 并体现出良好 的鲁棒性和自适应性 : 车辆工程; 操纵稳定性; 单神经元; 自适应PID 控制 : U270. 1 : A Single neuron adaptive control over vehicle steering performance 1 2 GAO Zhenhai , LI Xiangy ( 1. State Key Labor atory of A utomobile Dy namic Simulation , Jilin University , Changchun 130022, China; 2. College of A utomotive Eng ineering , J ilin University , Changchun 130022, China Abstract : Based on the total variance analysis on the vehicle steering stability and the ne ral netw ork control theory, an algorithm of the single ne ron adaptive PID control over the vehicle steering stability was proposed. Using the indices of trajectory follow error and steering b syness in eval ation of the steering stability, a q adratic form performance index f nction for the ne ron learning w as established and the t ning of the connection w eight val es of the single ne ron controller w as realized sing the algorithm of gradient descent. With vehicle nonlinear dynamic model of 7 degrees-of-freedom, the sim lations nder different speeds w ere made. The res lts show that this algorithm can control vehicle to follow the pre-given trajectory acc rately, and good rob stness and adaptability are also show n.