交通诱导系统中车流量预测与路径诱导算法研究

发布时间：2018-11-20 22:01

【摘要】：近年来,随着社会经济的发展,交通拥堵、交通事故等交通问题日益突出。为了应对这些问题,交通诱导系统被引入到城市的交通管理中,并且得到了快速的发展。其中车流量的短时预测与路径诱导是交通诱导系统的关键技术。对未来某时刻的车流量进行合理的预测,并给出合理的诱导路径,不仅能够为交通管理部门提供决策依据,而且能方便出行人出行,避免进入拥堵路段,节约出行时间。由于城市路网交通状态的时变性和复杂性,很难精确的描述其变化规律,因此研究实时准确的车流量预测与路径诱导算法具有十分重要的意义。本文通过对城市道路交通数据的分析,以城市道路网及交叉口为研究对象,对无检测器交叉口的车流量预测、路网车流量预测以及路径诱导算法进行了研究,论文主要研究工作包括以下几个方面:1.介绍了交通数据采集方法和交通数据的特性,分析了车流量预测的可行性,阐述了异常数据的识别与修复方法。采用时间序列分析法和Lyapunov指数分析并确定了车流量的可预测性,并使用历史趋势数据与实测数据的加权估计值对异常数据进行了修复。2.针对城市路网中某些交叉口没有检测器或者检测器故障的问题,在分析和研究几种常用无检测器交叉口车流量预测方法的基础上,提出了一种基于模糊C均值聚类的无检测器交叉口车流量预测方法。该方法通过模糊聚类将相关联的交叉口聚为同一簇,然后使用多元线性回归方法完成了对车流量的预测。实验结果验证了算法的有效性。3.通过对车流量预测模型的研究,给出了基于支持向量机回归方法的短时车流量预测模型,并针对SVR的参数学习速度慢的问题,研究了遗传算法的全局搜索特性,采用遗传算法优化SVR的参数选择,最后实验验证了GA-SVR模型的合理性。4.研究了几种传统的求解最优路径算法的原理,分析了它们的优缺点,在此基础上,引入了一种模拟进化的蚁群算法,对交通最优路径进行选择。该算法的主要原理是蚁群依靠与路径长度有关的信息素来寻找最优路径。同时针对蚁群算法的缺点对其进行改进,并用改进的蚁群算法与遗传算法进行实验对比分析,验证了算法的有效性。5.利用GA-SVR预测模型与蚁群最短路径诱导算法的研究结论,设计并完成了基于J2EE框架的交通诱导系统。
[Abstract]:In recent years, with the development of social economy, traffic congestion, traffic accidents and other traffic problems have become increasingly prominent. In order to deal with these problems, traffic guidance system has been introduced into urban traffic management and developed rapidly. Among them, the short-time prediction and route guidance of traffic flow are the key technologies of traffic guidance system. It can not only provide the decision basis for the traffic management department, but also facilitate the travel, avoid entering the congested section and save the travel time by reasonably forecasting the traffic flow at a certain time in the future. Because of the time variation and complexity of the traffic state of urban road network, it is difficult to describe its changing law accurately, so it is very important to study the real-time and accurate vehicle flow prediction and route guidance algorithm. Based on the analysis of urban road traffic data and taking the urban road network and intersection as the research object, this paper studies the vehicle flow prediction, road network traffic flow prediction and path guidance algorithm of the intersection without detector. The main research work includes the following aspects: 1. This paper introduces the methods of traffic data acquisition and the characteristics of traffic data, analyzes the feasibility of traffic flow prediction, and expounds the methods of identifying and repairing abnormal data. Time series analysis and Lyapunov index analysis are used to determine the predictability of traffic flow, and the weighted estimates of historical trend data and measured data are used to repair the abnormal data. 2. In view of the problem that some intersections in urban road network do not have detectors or fault detectors, based on the analysis and study of several commonly used traffic flow prediction methods of intersections without detectors, In this paper, a new method of traffic flow prediction at intersections without detector based on fuzzy C-means clustering is proposed. In this method, the associated intersections are clustered into the same cluster by fuzzy clustering, and the multivariate linear regression method is used to predict the traffic flow. The experimental results show that the algorithm is effective. Based on the research of traffic flow prediction model, a short-term traffic flow prediction model based on support vector machine regression method is presented. The global search characteristic of genetic algorithm is studied in view of the slow learning speed of parameters in SVR. Genetic algorithm is used to optimize the parameter selection of SVR. Finally, the rationality of GA-SVR model is verified by experiments. 4. 4. In this paper, the principles of several traditional optimal path algorithms are studied, and their advantages and disadvantages are analyzed. On this basis, a simulated evolutionary ant colony algorithm is introduced to select the optimal path of traffic. The main principle of the algorithm is that ant colony depends on the information related to path length to find the optimal path. At the same time, aiming at the shortcomings of ant colony algorithm, the improved ant colony algorithm and genetic algorithm are compared and analyzed, and the validity of the algorithm is verified. A traffic guidance system based on J2EE framework is designed and completed by using the GA-SVR prediction model and the conclusion of the ant colony shortest path guidance algorithm.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495

【参考文献】