【作者】 王春雷；

【导师】 郭孔辉；

【作者基本信息】 吉林大学 ， 车辆工程， 2017， 硕士

【摘要】 随着汽车的普及与发展,驾驶人数量逐年增加,其车辆驾驶水平呈现多样化发展趋势。面对相同的驾驶任务,具有不同驾驶经验或驾驶熟练度的驾驶员通常呈现不同车辆操控表现。如何在汽车的设计阶段就考虑广泛驾驶人员的车辆操纵特点,保证汽车具有出色的易驾驶性能,以此实现广泛驾驶员轻松、舒适以及出色的车辆驾驶表现是当前的研究热点之一。此外,对广泛驾驶人员车辆操纵行为规律的准确把握对未来汽车智能化、人性化发展亦具有重要意义。本文旨在建立一种可以表征广泛驾驶人员车辆转向操纵行为规律的数学模型,即考虑群体操纵特性的方向控制驾驶员模型(简称群体驾驶员转向控制模型),研究并揭示驾驶员群体操纵特性及其对车辆闭环跟随性能的影响,以此为车辆的适应人群能力评价提供了一种仿真分析手段。其中,驾驶员群体操纵特性的内涵解释为:广泛驾驶人员车辆驾驶过程中的操纵行为离散性与随机性,其离散性表征由驾驶员驾驶经验或驾驶熟练度不同的操纵行为差异性,随机性表征由于外界环境干扰与驾驶员自身生理心理偶然性等因素所引起的操纵行为不确定性。本文主要研究内容与结论如下:1.典型驾驶员转向控制行为建模。根据模型预测控制(简称MPC)理论对典型驾驶员的车辆转向控制行为,包括道路预瞄、方向盘转角决策以及方向盘操纵进行数学建模。在道路预瞄建模中,采用多点预瞄策略表征驾驶员观察前方道路获取期望路径信息的过程;在方向盘转角决策建模中,引入以等效侧偏刚度为变量的线性二自由度车辆模型组表征驾驶员对于不同车辆动态特性的认知,并以此作为预测模型,制定了相应MPC转向控制规则;在方向盘操纵建模中,考虑了驾驶员神经肌肉系统的反应滞后效应并采用时间延迟环节表征该延迟特性。通过典型驾驶员转向控制模型研究,揭示了驾驶员车辆转向任务中预瞄、决策以及执行等行为特性,为考虑群体操纵特性的方向控制驾驶员建模奠定基础。2.典型驾驶员转向控制模型求解。依据状态空间法建立典型驾驶员转向控制模型的解析与数值求解方法。在解析法中,通过极值点求导获得最优解析方向盘转角;在数值法中,建立了表征真实驾驶员生理限制等因素的约束条件,并采用一种原对偶-有效集二次规划算法获得最优数值方向盘转角。通过两种算法在不同工况下的对比仿真分析,验证了所述典型驾驶员转向控制模型的有效性,且明确了该模型数值解法在求解精度与速度方面的优越性,为群体驾驶员转向控制模型解算奠定基础。3.群体驾驶员转向控制行为建模。以典型驾驶员转向控制模型为基础,建立刻画广泛驾驶人员操纵行为离散性与随机性的数学模型。其中,以驾驶员对车辆状态的认知特性为切入点,提出一种轮胎侧偏角感知模型。该模型中,驾驶员对车辆侧向动力学的认知行为被具体描述为一种对轮胎侧偏角的感知行为,其感知轮胎侧偏角为服从高斯分布的随机变量,均值对应当前采样时刻的真实轮胎侧偏角,方差表征驾驶员驾驶熟练度与车辆侧向动力学非线性特性。最后,利用真实驾驶员双移线车辆轨道数据对上述轮胎侧偏角模型参数进行了辨识。4.群体驾驶员转向控制模型验证与分析。利用Car Sim-MATLAB/Simulink联合仿真手段,通过不同双移线工况下仿真车辆轨道数据库与真实车辆轨道数据库包络区域、车辆轨道误差统计特征的对比分析,验证了群体驾驶员转向控制模型的有效性,并据此对熟练驾驶人群、一般驾驶人群的车辆转向操纵行为规律进行分析。结果表明,在执行某一相同车辆轨道跟踪任务时,熟练驾驶人群相比一般驾驶人群表现更加精确且稳定;此外,驾驶员车辆轨道跟踪表现会随着车辆侧向动力学非线性的增加而降低。5.基于群体驾驶员转向控制模型的车辆易驾驶性能分析,即依据所提出的群体驾驶员转向控制模型,探索汽车操纵性能适应驾驶人群能力的仿真分析方法。首先,从驾驶安全与车辆闭环响应两方面构建了人车闭环性能评价指标;其次,依据人车闭环仿真试验建立了人车闭环性能评价指标数据库,并通过闭环性能评价指标数据的概率分布与相关统计特性(均值与方差)分析明确了同一车辆其不足转向适应人群能力相较于过多转向适应人群能力更强;最后,通过对车辆质心至后轴距离这一参数进行优化实现了车辆在不足转向特性条件下的最佳适应人群表现。通过本文研究,实现了典型驾驶员车辆转向任务中预瞄、决策以及执行等行为特性的准确刻画,建立了表征广泛驾驶人员操纵行为离散性与随机性的数学模型,为数字化、智能化以及人性化的汽车设计开发奠定理论基础。更多还原

【Abstract】 With the popularization and development of vehicle techniques,the total number of the drivers has been gradually increasing,and the driver’s driving skills are becoming more and more diversified.Considering the same driving task,drivers with the different driving skills usually present different vehicle manipulation performance.In this way,how to consider the wide driving crowd’s driving behaviours within the initial vehicle design process,providing an good vehicle ease-of-control performance is a hot research topic for researchers in recent years.Moreover,an accurate grasping of the wide driving crowd’s driving behaviours is also important for the intelligent and personalized development of the vehicle.The paper aims at establishing an mathematical model being capable of characterizing the wide driving crowd’s steering behaviour laws,i.e.,the steering control driver model with the driver group’s steering maneuvers,studying the steering control laws of the wide driving crowd as well as its influences on the closed-loop path-following performances,to provide a simulation method using for the adaptability analysis of vehicles to drivers.In particular,the "driver group manipulation laws" can be explained as the discreteness as well as randomness of the wide driving crowd’s vehicle maneuvers: the discreteness characterizes the differences of maneuver behaviours due to drivers’ various driving experiences or skills;the randomness characterizes the uncertainty of behaviors owning to the external environmental interferences,drivers’ physiological and psychological factors,and so on.The main contents and conclusions are as follows:1.Modeling on typical driver steering control behaviors.Based on the Model Predictive Control(MPC)theory,three typical steering control maneuvers of the drivers,including the road preview,steering wheel angle decision and steering wheel manipulation are modeled.In the road preview modeling process,a multi-point previewing strategy is used to characterize the driver’s behavior of obtaining the desired path information from road ahead.Within the steering angle decision modeling process,the linear two DOF vehicle model groups with the equivalent cornering stiffness as variables are proposed to characterize the drivers’ cognition behavior to the different vehicle dynamic characteristic,which are used as predictive models,and then MPC steering control rules are derived.Within the modeling process of the steering wheel manipulation,the hysteresis effect of the driver’s neuromuscular system is modeled by a time delay element.Through modeling on typical driver steering manipulation,the typical driving behaviors,including the preview,decision and the execution of a driver are revealed,which lays foundations for study on the steering control driver model with the driver group’s steering maneuvers.2.Solution to the typical driver steering control model.In accordance to the state space method,two solving strategies,including the analytical and numerical ones are used to solve the typical driver steering control models.During the analytical solving process,the optimal steering wheel angle is obtained via derivation.As for the numerical solution,the constraints featuring the real drivers’ physiological restrictions are established,and a primal-dual active set quadratic programming method is utilized to obtain the optimal steering wheel angle.The proposed model is validated via simulation and the comparison results with the two methods under different driving conditions indicate the superior precision and speed of the numerical solution.3.Modeling on driver group steering control behaviors.Based on the previously derived typical driver steering control model,the steering maneuver considering the discreteness and randomness of the wide driving crowd is modeled.Wherein,the driver’s cognition behaviors to the vehicle states are investigated and a tire side slip angle perception model is proposed.For this model,driver’s cognition to vehicle lateral dynamics are described as the perception to the tire side slip angle,whose perceived side slip angle is a random variable subject to the Gaussian distribution,the mean of which is corresponding to the real side slip angle and the variance characterizes the driver’s steering skills and the nonlinear vehicle lateral dynamics.Eventually,the parameters of the tire side-roll angle model are identified via the real driver’s double-lane change vehicle trajectory data.4.Validation and discussion on the group steering control model.To verify the proposed group driver model,a Car Sim-MATLAB/Simulink co-simulation platform is built.Through the comparisons of the simulated and the real vehicle trajectory database envelope regions,the statistics of the vehicle trajectory errors,the results show that the proposed group driver steering control model is accurate.Based on this model,the steering manipulation laws of the skilled and novice driving crowds are discussed.The results manifest that faced with the same path tracking tasks,the skilled driver is more accurate and stable than the novice driver,and the tracking performances will decrease with the increase of the vehicle lateral dynamics nonlinearity.5.Vehicle ease-of-control evaluations.With the group driver steering control model,the simulation method for vehicle ease-of-control evaluation is studied.Wherein,a performance evaluation criterion with considerations of driving safety and vehicle closed-loop response is established.Through the simulation tests,a database of the performance evaluation criterion is built.Based on the statistics of the criterion data,more excellent ease-of-control evaluation results for vehicles with understeering characteristic are determined,and the rear wheelbase is optimized to realize the best performance of vehicle with the understeering characteristic.In this paper,the behavioral characteristics of the typical drivers’,including the preview,decision making and the execution in the steering task is accurately modeled,furthermore,a mathematical model is derived to characterize the discreteness and randomness of the wide driving crowd,which lays a theoretical foundation for the digital,intelligent and personalized development of the vehicle.更多还原