两点预瞄驾驶员转向模型研究

发布时间：2018-04-22 07:31

本文选题：驾驶员模型 + 转向行为　；参考：《南京航空航天大学》2016年硕士论文

【摘要】：随着人们生活水平的不断提高,汽车的保有量不断增加,随之而来的交通安全问题越来越引人注目,人们对汽车安全性能的要求也越来越高。因此,如何对汽车安全性能进行客观、公正的评价,成为了研究人员急需解决的问题。然而由于许多实车试验都是在极限工况下进行,对驾驶员的生命安全问题构成了严重威胁,为此本文对驾驶员的转向操纵行为进行研究,以期建立一个与真实的驾驶员转向操纵行为具有良好一致性的驾驶员转向模型取代真实的驾驶员,对汽车安全性能进行客观、公正的评价。另外驾驶员转向模型的研究对智能车转向系统的开发也具有一定的参考价值。本文以驾驶员转向模型为研究对象,结合车辆模型、道路模型建立人-车-路闭环控制系统模型,其主要的工作内容如下:首先,建立车-路模型。第一步对车辆的横向运动进行受力分析,建立线性二自由度车辆动力学模型;第二步将离散的道路数据点进行曲线拟合,建立S形曲线道路模型;第三步基于车辆模型和道路模型建立车、路相对位置模型,即车-路模型。随后,建立驾驶员子系统模型。第一步对驾驶员转向操纵行为进行整体分析,探明在车辆转向行驶过程中,有哪些驾驶员子系统参与其中;第二步借鉴人类科学相关研究成果对参与转向操纵行为的驾驶员子系统进行分析与数学建模,最终建立了基于两点预瞄视觉机制的视觉子模型、前庭横摆加速度感知子模型、基于大脑情感学习回路(BEL)的转向决策子模型、手臂神经肌肉子模型。最后,建立人-车-路闭环控制系统模型,并进行相关仿真分析。本部分要解决的重点问题是将前两部分建立的车-路模型以及驾驶员子模型进行有机结合,形成人-车-路闭环控制系统模型。第一步建立基于BEL模型的单点预瞄驾驶员转向模型,并将其与单点预瞄最优曲率驾驶员转向模型进行对比分析,验证了BEL驾驶员决策模型的有效性以及该模型与真实驾驶员转向决策行为的一致性;第二步在基于BEL模型的单点预瞄驾驶员转向模型的基础之上建立两点预瞄驾驶员转向模型,并将其与基于BEL模型的单点预瞄驾驶员转向模型进行对比分析,验证了两点预瞄视觉机制优于单点预瞄视觉机制;第三步对基于BEL模型的两点预瞄驾驶员转向模型与基于LQR的多点预瞄驾驶员转向模型进行对比分析,分析结果表明在跟踪特性上多点预瞄驾驶员转向模型的性能优于两点预瞄驾驶员转向模型,但两点预瞄驾驶员转向模型也能够很好的完成目标道路轨迹的精确跟踪,并且各项性能指标优越,即两者之间在跟踪性能上差距不大。综上所述,由于本文所建立的驾驶员转向模型是基于对真实驾驶员转向行为分析基础之上所建立的,因此其物理意义清晰,能够很好地再现真实驾驶员的转向行为,与现实中驾驶员的转向行为具有良好一致性。另外该模型能够很好的实现目标道路轨迹的精确跟踪,对智能车转向系统的开发具有一定的参考价值。
[Abstract]:With the continuous improvement of people's living standards, the number of cars kept increasing, the attendant traffic safety problems are becoming more and more noticeable, and the demand for car safety performance is becoming more and more high. Therefore, how to conduct objective and fair evaluation of automobile safety performance has become an urgent problem for the researchers to solve. The test of multi real car is carried out under the extreme condition, which poses a serious threat to the driver's life safety problem. Therefore, this paper studies the steering control behavior of the driver in order to establish a driver steering model which is in good agreement with the real driver's steering control behavior to replace the real driver and to the automobile safety. In addition, the research of driver steering model also has certain reference value for the development of intelligent vehicle steering system. This paper takes the driver steering model as the research object, combines the vehicle model and the road model to establish a man car road closed loop control system model. The main work contents are as follows: first, build The first step is to analyze the lateral motion of the vehicle and establish a linear two degree of freedom vehicle dynamic model. The second step is to fit the discrete road data points and establish the S shape curve road model; the third step is based on the vehicle model and road model, the road relative position model, the car road model. Then, the vehicle model and the road model are established. The first step is to establish the driver's subsystem model. The first step is to analyze the driver's steering control behavior as a whole, and to find out what driver subsystem is involved in the steering of the vehicle, and the second step is to analyze and model the driver subsystem of the steering control by using the related research results of human science, and finally establish the model of the driver's subsystem. The visual submodel based on the two-point preview visual mechanism, the vestibule yaw acceleration perception sub model, the steering decision submodel of the brain affective learning loop (BEL) and the arm neuromuscular model. Finally, the model of the human vehicle road closed loop control system is established and the related simulation analysis is carried out. The key problem to be solved in this part is the first two. In the first step, a single point preview driver steering model based on BEL model is established, and the model is compared with the single point preview driver model of the optimal curvature driver, which validates the effectiveness of the BEL driver decision model. As well as the consistency of the model and the real driver's turning to the decision-making behavior, the second step is based on the single point preview driver steering model based on the BEL model, and establishes a two point preview driver steering model, and compares it with the single point preview driver steering model based on the BEL model, and verifies the two preview vision machine. The system is superior to the single point preview visual mechanism; the third step based on the BEL model is compared with the LQR based multi point preview driver steering model. The analysis results show that the performance of the multi point preview driver steering model is better than the two point preview driver steering model on the tracking characteristics, but the two points preview is preaimed. The driver's steering model can also accurately track the track track of the target, and the performance index is superior, that is, there is little difference in the tracking performance between the two. In summary, the driver steering model is based on the analysis of the real driver's steering behavior. The meaning is clear and the steering behavior of the real driver can be reproduced well. It is in good agreement with the driver's steering behavior in reality. In addition, the model can accurately track the path of the target Road, and has certain reference value for the development of the intelligent vehicle steering system.

【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U463.4

【参考文献】