南昌市智能交通流量预测算法研究

发布时间：2018-06-02 12:54

本文选题：智能交通 + 交通流量　；参考：《南昌大学》2015年硕士论文

【摘要】：大量的交通流数据隐藏在城市交通信息中,这些数据潜在的蕴含了交通流现象和一些意想不到的规律,能全面、科学和反映各街道交通的实际情况,是实现科学化交通管理、制定科学化交通决策必不可少的依据,是十分宝贵和重要的信息资源,也是居民出行了解路况的主要参考依据之一。由于城市交通流信息中包含着大量可挖掘出的有效的交通数据信息,能为实际交通流量预测起关键性作用,所以采取何种方法去挖掘其隐含的有实际利用价值的信息是当今社会的重要研究热点。本文以南昌市东湖区历史真实数据的道路交通数据为依据,以相应交通流量预测算法研究为主旨,对获取到的东湖区实际交通流数据进行分析,然后在对相应预测算法的了解和认识,进行交通流量的预测探讨和研究。具体研究方案是,先清洗获取到的交通流数据,用程序统计出某个街道某一天的5分钟为单位的交通流量值,然后采用聚类算法中的K-means算法先对这些数据根据流量的值分为4类,然后再在各个类中,按照时间的大小来聚为2、3、3、2类,一共20个类,最后再针对各个类,利用最小乘支持向量机预测模型来进行各类时间段的流量预测。同时论文也进行了单独最小乘支持向量机预测模型来预测,然后将其结果和组合算法模型的结果相比较,得出组合算法的科学性有效性结论。
[Abstract]:A large number of traffic flow data are hidden in urban traffic information. These data potentially contain traffic flow phenomena and some unexpected laws, which can comprehensively, scientifically and reflect the actual situation of each street traffic, and realize scientific traffic management. Making scientific traffic decision making is an indispensable basis, which is a very valuable and important information resource, and is also one of the main reference bases for residents to get to know the road conditions. Because the urban traffic flow information contains a lot of effective traffic data information which can be excavated, it can play a key role in the actual traffic flow prediction. Therefore, it is an important research hotspot to explore the hidden information with practical value. Based on the road traffic data from the real historical data of Donghu District, Nanchang City, this paper analyzes the actual traffic flow data obtained in the East Lake region, based on the research of the corresponding traffic flow forecasting algorithm. Then, the traffic flow prediction is discussed and studied in the understanding and understanding of the corresponding prediction algorithm. The specific research plan is to first clean the traffic flow data obtained, and then use the program to calculate the traffic flow values per unit of 5 minutes of a certain street on a certain day. Then the K-means algorithm of clustering algorithm is used to divide these data into four categories according to the value of traffic, and then in each class, according to the size of the time, the data is grouped into two classes, 20 classes, and finally for each class. The prediction model of minimum multiplication support vector machine is used to predict the flow of all kinds of time periods. At the same time, the prediction model of single least multiplication support vector machine is carried out, and then the results are compared with the results of the combined algorithm model, and the scientific and effective conclusion of the combined algorithm is obtained.
【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491.14

【参考文献】