驾驶员对汽车方向的自适应控制行为建模  

Modeling of Driver's Adaptive Control Behavior for Vehicle Direction

作　　者：段立飞[1] 高振海[1] 王德平[2] 

机构地区：[1]吉林大学汽车动态模拟国家重点实验室,长春130025 [2]中国第一汽车集团公司,长春130011

出　　处：《机械工程学报》2011年第8期121-125,133,共6页Journal of Mechanical Engineering

基　　金：国家自然科学基金(50975120;90820014);国家高技术研究发展计划(863计划;2006AA110102);霍英东基金(114025);教育部科学技术研究重点项目(108041)和教育部优秀新世纪人才支持计划(NCET-08-0247)资助项目

摘　　要：模拟驾驶员在驾驶学习过程中对汽车动力学特性的学习行为,建立汽车驾驶员对汽车方向的自适应控制行为模型。根据驾驶员对汽车方向控制的熟练程度,将驾驶员分为合格驾驶员和专业驾驶员。在驾驶学习过程中,新手驾驶员经过低速缓慢转向工况的反复学习和训练后对汽车动力学特性有基本的了解,可以在低速线性区内熟练地控制汽车。考虑低速区内驾驶员控制行为的特点,建立基于遗传算法离线整定的方向控制模型模拟新手驾驶员成为合格驾驶员的过程。当车速较高或侧向加速度很大时,车辆动态响应具有明显的强非线性特性,合格驾驶员需要经过高速驾驶经验的累积过程才能熟练地控制汽车,成为一名专业驾驶员。因此,针对高速行驶时汽车的非线性动力学特性,在原有模型基础上引入神经网络在线整定的方法,模拟合格驾驶员经高速行驶训练成为专业驾驶员的学习过程。模型的仿真结果与真实驾驶员操纵行为具有很好的一致性。An adaptive directional control driver model is established to simulate the driver＇s learning behavior about vehicle dynamic characteristic.Drivers are divided into eligible group and professional group according to their directional control skill.During the learning process,novice drivers will have a basic understanding of vehicle dynamics after low speed slow steering training,so they can proficiently control the vehicle at low speed range.A direction control model based on genetic algorithm offline tuning is built to simulate this learning process from a novice driver to an eligible driver in light of the characteristics of driver＇s control behavior at low speed range.The dynamic response of vehicle shows a strong nonlinear property when the lateral acceleration or the speed is high,eligible drivers need to gather the driving experience at high speed range to control the vehicle skillfully,finally becoming an expert driver.So,for strong nonlinear characteristic of the vehicle at high speed,neural network online tuning is added to the original model to simulate learning process from an eligible driver to a professional driver under high-speed condition.The simulation results of model are well consistent with the behavior of real driver.

关 键 词：驾驶员模型 高速非线性 方向自适应控制 遗传算法 神经网络 

分 类 号：U471.3[机械工程—车辆工程]