【作者】 张鑫；

【导师】 谢辉； 岳攀；

【作者基本信息】 天津大学 ， 工程（专业学位）， 2021， 硕士

【摘要】 随着我国汽车运输行业的快速发展,我国公路货运卡车数目逐渐增多,但是驾驶员的驾驶水平参差不齐,卡车驾驶多为长途运输,会造成驾驶员急加速、急刹车、超速、不按规定行驶等不良驾驶习惯。驾驶员是车辆的操纵者,是道路交通的主要参与者,因此对驾驶员驾驶行为分析及评价研究驾驶改善经济性和安全性,具有重要意义。本研究建立了驾驶行为分析系统,提出了一套驾驶经济性与驾驶安全性的综合指标体系和和驾驶行为的评价方法。研究的主要内容如下:以某企业货运卡车为研究对象,对超过6万公里的实际车辆运行数据进行数据分析与处理。并且开发了车辆的里程数据重构,车重信息估计,缺失、错误等数据的重构算法,提取数据特征,用于描述刻画驾驶员的驾驶行为。根据驾驶风格对发动机工况的影响,通过机理分析,得到发动机燃油消耗与驾驶员踏板开度,发动机转速,扭矩,车速的关系,并通过实际数据分析以及皮尔逊相关系数进行验证,提炼出影响油耗的驾驶行为特征参数,为经济性评价模型建立奠定基础。然后将所需数据进行数据标准化处理,根据特征参数以瞬时油耗为导向采用聚类算法将驾驶行为进行分析并打上各个等级的标签,以特征参数指标为输入,以做了标签的瞬时油耗为输出建立基于神经网络的驾驶经济性评价模型。其次,对不同驾驶员采用控制变量的方法进行对比分析,以车速、加速度、加速度的变化率为指标将安全性分为三个等级,结合层次分析法确定各等级的指标权重,统计出各个驾驶员的安全性驾驶水平。最后结合开发的驾驶行为评价系统,对驾驶员的行为特征进行画像,将驾驶员的基本属性、驾驶行为特征,特征在群体中的排名,变化趋势和评估得分结合在一起,形成驾驶行为分析评价系统,可以用来综合评估驾驶员驾驶习惯和操纵行为的优劣、驾驶行为的优劣,以此为基准可以用来改善司机的驾驶行为,优化驾驶操控习惯。更多还原

【Abstract】 With the rapid development of China’s automobile transportation industry,the number of highway freight trucks in China is gradually increasing,but the driver’s driving level is uneven,truck driving for long-distance transportation,will cause drivers to accelerate,brake,speed,do not drive according to the provisions and other bad driving habits.The driver is the operator of the vehicle and the main participant of the road traffic,so it is of great significance to analyze the driver’s driving behavior and evaluate the study on improving the economy and safety of driving.Based on the driving behavior analysis system of this study,a set of driving economy and driving safety index system and evaluation methods are proposed.The main contents of the study are as follows:Taking the freight truck of an enterprise as the research object,the actual vehicle running data of more than 60,000 kilometers is analyzed and processed.And developed the reconstruction algorithm of vehicle mileage data reconstruction,vehicle weight information estimation,missing,error and other data,extracted data features,used to describe the driver’s driving behavior.According to the driving style’s influence on the engine operating conditions,through the mechanism analysis,the engine fuel consumption and the driver pedal opening,engine speed,torque,speed,and validated through actual data analysis and Pearson correlation coefficient,extracting affect driving behavior characteristic parameters of fuel consumption,lay a foundation for the economic evaluation model.Then,the required data were standardized.According to the characteristic parameters,the instantaneous fuel consumption was taken as the guide and the clustering algorithm was adopted to analyze the driving behavior and label each level.The driving economy evaluation model based on neural network was established with the characteristic parameter index as the input and the instantaneous fuel consumption with the label as the output.Secondly,the method of using control variables for different drivers is compared and analyzed.The safety level is divided into three levels with the speed,acceleration and the rate of change of acceleration as indicators.The index weight of each level is determined with the analytic hierarchy process(AHP)to calculate the safety driving level of each driver.Finally combined with the development of driving behavior evaluation system,to portrait of driver behavior characteristics,basic attribute of drivers,driving characteristics,the characteristics of the rankings in groups,change trend and evaluation score together,form the driving behavior evaluation system,which can be used to comprehensive evaluate driver’s driving habits and manipulation,driving economy and driving safety,as a benchmark used to improve the driver’s driving behavior,optimize the driving manipulation of the habit.更多还原