基于Elman神经网络的赛车操纵动力学逆向研究  

Inverse Research on Formula Racing Car Handling Dynamics Based on Elman Neural Network

作　　者：倪俊[1] 陈思忠[1] 吴志成[1] 

机构地区：[1]北京理工大学机械与车辆学院,北京100081

出　　处：《机械设计与研究》2012年第4期105-107,112,共4页Machine Design And Research

基　　金：国家部委预研基金资助项目(C2220061355)

摘　　要：以进行汽车操纵动力学逆向研究为目的。以某方程式赛车为例,建立了计入空气动力特性的多体动力学整车模型.在通过试验验证其准确度的基础上,利用Elman神经网络建立了单移线工况下的侧向加速度、横摆角速度与方向盘转角之间的非线性映射关系.神经网络的验证结果表明,利用Elman神经网络进行汽车操纵逆动力学的研究方法是可行的,并且学习速率较快,识别精度较高。In order to research on vehicle inverse handling dynamics, a certain formula racing car is taken as an example. The mulit-body model with aerodynamic characteristics was built. Based on its correctness verified by experi- ment data, the nonlinear mapping relation among lateral acceleration, roll angular velocity and steering angle was built by Elman neural network. The indentification result shows that the method for studying the vehicle inverse handling dy- namics by Elman neural network is feasible, which has a rapid learning speed and high accuracy.

关 键 词：方程式赛车 操纵动力学 逆问题 ADAMS ELMAN神经网络 

分 类 号：U461.6[机械工程—车辆工程]