面向排放测算的Wiedemann和Fritzsche跟驰模型优化研究

发布时间：2019-04-27 07:55

【摘要】：摘要：随着我国城镇化的加速发展,城市机动车保有量急剧增长,机动车尾气污染物排放日益成为影响人们日常生活的主要问题之一。微观交通仿真模型作为衔接城市交通流运行评价与机动车尾气污染物测算的重要手段,其对城市机动车污染物排放评价的功能越来越受到众多学者的应用与研究。然而,由于内部仿真机理存在系统缺陷,微观交通仿真模型的核心模块,车辆跟驰模型的排放测算精度尚不能满足排放测算要求。为此,本文以机动车排放测算的最佳解释变量VSP分布作为研究基础,重点分析与探讨了提高Wiedemann模型与Fritzsche模型排放测算精度的方法与途径,并对优化效果进行验证。首先,对生理-心理跟驰模型进行重点分析,研究Wiedemann模型和Fritzsche模型的内部跟驰逻辑与仿真机理。此外,对国内外的微观排放模型和面向排放测算的车辆跟驰模型优化研究展开综述与分析。其次,利用便携式车载GPS展开数据收集和仿真方法设计。本文利用便携式车载GPS收集了大量的北京市快速路上实际驾驶行为数据和典型的跟驰驾驶行为数据,利用收集到的实际驾驶行为数据和实测跟驰驾驶行为数据,本文根据相同道路等级下的驾驶行为数据具有相同的分布特征,研究设计了Wiedemann模型和Fritzsche模型的Matlab数值仿真方法。在此基础上,基于仿真得出的Wiedemann模型和Fritzsche模型在各个跟驰状态下的比例,对Wiedemann模型的速度跟驰阈值SDV、CLDV以及距离跟驰阈值SDX展开相关敏感性分析。根据敏感性分析结果,重点对距离跟驰阈值SDX展开优化研究,并提出了跟驰域SDX改进的Wiedemann模型。此外,基于本文实际采集的大量驾驶行为数据,拟合建模车辆行驶最大加速度与瞬时速度的关系式,并以此作为Wiedemann模型和Fritzsche模型最大加速度的改进关系式。经过模型整体优化效果分析,Wiedemann模型仿真输出的VSP分布在各个速度区间下的平均相对均方根误差,从原来的0.565减低到0.102,降幅明显；改进后的Fritzsche模型相比于原模型,其仿真输出的VSP分布在各个速度区间下的平均相对均方根误差降低了57.23%。最后,以实测跟驰驾驶行为数据为基础,设计了车辆跟驰模型的模型稳定性控制算法,并分别从车头时距稳定性、加速度合理性和仿真速度RMSE等角度对改进后的Wiedemann模型和Fritzsche模型进行模型稳定性验证。结果表明,改进后的模型仍具备较高的稳定性。
[Abstract]:Abstract: with the rapid development of urbanization in China, the number of urban motor vehicle ownership has increased dramatically, and the emission of vehicle exhaust pollutants has become one of the main problems affecting people's daily life. As an important means of linking urban traffic flow evaluation with vehicle exhaust pollutant estimation, microscopic traffic simulation model has been applied and studied more and more by many scholars in the evaluation of urban vehicle pollutant emission. However, due to the defects of the internal simulation mechanism, the emission measurement accuracy of the vehicle-following model, the core module of the microscopic traffic simulation model, can not meet the requirements of emission measurement. Therefore, based on the VSP distribution of the best explanatory variable of vehicle emission measurement, this paper analyzes and discusses the methods and ways to improve the precision of Wiedemann model and Fritzsche model, and validates the optimization effect. Firstly, the physiological-psychological following model is analyzed, and the internal following logic and simulation mechanism of Wiedemann model and Fritzsche model are studied. In addition, domestic and foreign microscopic emission models and vehicle-following model optimization for emission measurement are reviewed and analyzed. Secondly, the method of data collection and simulation is designed by using portable vehicle GPS. In this paper, a large number of actual driving behavior data and typical following driving behavior data on Beijing expressway are collected by portable vehicle GPS, and the actual driving behavior data and measured driving behavior data are used. In this paper, according to the same distribution characteristics of driving behavior data under the same road grade, the Matlab numerical simulation method of Wiedemann model and Fritzsche model is studied and designed. On this basis, based on the proportions of the Wiedemann model and the Fritzsche model under each following state, the sensitivity analysis of the velocity-following threshold SDV,CLDV and the distance-following threshold SDX of the Wiedemann model is carried out. According to the results of sensitivity analysis, the optimization of distance-following threshold (SDX) is focused on, and an improved Wiedemann model based on SDX in the range-following region is proposed. In addition, based on a large number of driving behavior data collected in this paper, the relationship between the maximum acceleration and the instantaneous velocity of the vehicle is modeled and used as an improved formula for the maximum acceleration of the Wiedemann model and the Fritzsche model. Through the analysis of the whole optimization effect of the model, the average relative root mean square error of the VSP distribution of the simulation output of the Wiedemann model in each velocity range is reduced from 0.565 to 0.102, and the decrease is obvious. Compared with the original model, the average relative root mean square error of the VSP distribution of the improved Fritzsche model is reduced by 57.23% in each velocity range. Finally, based on the measured driving behavior data, the model stability control algorithm of the vehicle following model is designed, and the stability of the vehicle following model is obtained from the time-distance between the front of the car and the vehicle. The stability of the improved Wiedemann model and Fritzsche model is verified from the angle of acceleration rationality and simulation velocity RMSE. The results show that the improved model still has high stability.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491

【参考文献】