基于GRNN网络车辆方向盘转角识别方法  

Vehicle Steering Angle Identification based on GRNN

作　　者：武健[1,2] 赵又群 邱笑寅 

机构地区：[1]南京航空航天大能源与动力学院车辆工程系,江苏南京210016 [2]聊城大学机械与汽车工程学院,山东聊城252059

出　　处：《应用基础与工程科学学报》2014年第1期170-178,共9页Journal of Basic Science and Engineering

基　　金：国家自然科学基金资助项目(11072106)

摘　　要：为改善车辆的操纵安全性以及为自动转向系统和智能泊车系统的研究提供理论基础,本文给出一种新的汽车方向盘转角识别方法.建立了驾驶员—车辆闭环模型,并通过试验验证车辆模型.通过给定方向盘转角输入求解得到用于GRNN网络训练的车辆状态参数,运用GRNN网络建立以车辆状态参数来识别方向盘转角的映射模型.与RBF网络相比,GRNN神经网络具有更高的辨识精度.进行整车仿真,所建立的GRNN神经网络能较精确地识别方向盘转角并与仿真结果有较好的一致性.The identification of steering wheel angle belongs to vehicle handling inverse dynamics problems. A new method based generalized regression neural network was proposed to identify the steering wheel angle, and a driver-vehicle model was built and verified by real vehicle tests. MATLAB was used to solve the problem normally to get generalized regression neural network training samples, then the generalized regression neural network was used to establish a nonlinear mapping model using vehicle state parameters to identify the steering wheel angle. Generalized regression neural network model has better robustness and interference immunity. Compared with radial basis function network, generalized regression neural network has higher recognition accuracy. Vehicle simulation is conducted by the ADAMS/CAR module, which shows that the generalized regression neural network can establish a more precise identification for steering wheel angle than radial basis function network, and is coherent well with the ADAMS/CAR simulation.

关 键 词：自动驾驶 方向盘转角识别 人-车闭环模型 GRNN 

分 类 号：U461[机械工程—车辆工程]