基于信号周期的间断流行程时间短时预测

发布时间：2018-02-02 19:42

本文关键词： 间断流信号周期路段行程时间马尔科夫链样条权函数　出处：《北方工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：人们的便利出行需要智能交通系统(Intelligent Transportation Systems,ITS)引导与控制。而依据信号交叉口划分的路段是城市道路最基本的构成元素,也是ITS系统最基本的数据信息来源和管控单元。路段行程时间也最能表征交通拥堵状况,实时体现的道路交通拥堵程度,在交通规划、交通管控中起着重要作用。出行者、交通管理者及研究者都经常用其评价交通状况或作为下一步交通决策的依据。本文将城市道路信号交叉口交通流描述为车辆随机到达的间断流,基于信号周期对路段行程时间进行估计和预测。全文主要研究内容和成果如下:(1)首先对路段交通流进行特性分析,根据特性不同将路段细分为交叉口域和路段域。针对路段域交通流为不受信号周期影响的平稳时间序列连续流,采用ARIMA算法建立路段域行程时间估计模型。通过对采集的路段域GPS数据进行地图匹配和位置时间插值方法计算的实例分析,结果表明基于ARIMA算法能准确地对路段域行程时间进行估计。(2)根据车辆在信号周期的红黄灯时间还是绿灯时间到达交叉口,对交叉口域行程时间进行了再划分,提出了基于随机车辆到达的马尔科夫链排队论和交通流波动理论的排队模型,并根据交叉口信号周期配时方案,建立交叉口域行程时间估计模型。实例分析结果显示该模型能较好地估计交叉口域行程时间。(3)在以上研究工作的基础上,基于样条权函数神经网络模型,并考虑和分析行程时间的关键影响因素,最后将路段域行程时间、交叉口域行程时间和信号周期、有效绿灯时长、绿信比作为行程时间影响因素,建立了行程时间预测模型。经过实例分析并和BP神经网络、RBF神经网络和卡尔曼滤波比较,最后选择路段域行程时间、交叉口域行程时间和信号周期为关键影响因素的样条权函数神经网络模型能更好地预测实验路段行程时间。
[Abstract]:The convenience of people travel needs of intelligent traffic system (Intelligent Transportation Systems, ITS) guide and control. And based on the signal intersection Road city road is the most basic element, is the most basic ITS system data sources of information and control unit. The road travel time also can best represent the traffic congestion, road traffic congestion degree the real-time reflect, in traffic planning, plays an important role in traffic control. Travelers, and traffic management research are often used in the evaluation of traffic or traffic as the next decision basis. This will be the city intersection traffic flow is described as vehicles were reached between the cutoff, signal period estimation and for the prediction of travel time. Based on the main research contents and results are as follows: (1) the traffic flow characteristics were analyzed, according to the different characteristics of the road Segment for intersection and road section. According to the domain domain domain traffic flow is a stationary time series of continuous flow without signal cycles, using ARIMA algorithm to establish the link travel time estimation model. Domain analysis calculation of map matching and location time interpolation method by road domain GPS the collecting data, results show that the ARIMA algorithm can based on accurately on the road travel time domain estimation. (2) according to the time of yellow light red or green light vehicles in the signal cycle time arrived at the intersection, the intersection of travel time domain were then divided, put forward the queuing model of Markoff chain stochastic vehicle queueing theory and traffic flow based on the wave theory, and according to the intersection signal timing scheme, establishing domain intersection travel time estimation model. The results show that the model can estimate the intersection travel time domain better. (3 ) on the basis of the above study, the spline weight function based on neural network model, and considering the key factors and effect analysis of the travel time, the road travel time domain, domain of travel time and intersection signal cycle, effective green time, green ratio as the travel time factors, travel time prediction model is established. Through the example analysis and BP neural network, RBF neural network and Calman filter, the final choice of domain road travel time, intersection travel time domain and signal cycle is the key impact factors of spline weight function neural network model can better predict the travel time.

【学位授予单位】：北方工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491.14

【参考文献】