城市道路交通流预测与路径选择研究

发布时间：2018-04-20 11:45

本文选题：K-Means算法 + 交通流量预测　；参考：《浙江工业大学》2015年硕士论文

【摘要】：随着城市汽车保有量的不断增长,车辆尾气污染和道路交通拥堵问题愈发严重。交通诱导系统可以有效地减少车辆污染和缓解交通拥堵。准确预测城市道路交通流量对交通诱导起到非常关键的作用,合理地路径优化可以为交通诱导提供方案,从而达到解决交通问题的目的。为了提高交通流量预测的准确率,本文提出了在交通流数据聚类分析之后,再采用神经网络进行预测的方法。本文的主要研究内容和成果如下:(1)分析了交通流量预测技术和路径选择技术的原理、研究现状与应用。(2)针对交通流数据的特性和影响因素,提出了一种基于K-Means算法和遗传算法(GA)优化的小波神经网络(WNN)的交通流量预测方法。鉴于K-Means算法能够快速高效地处理大规模数值型数据的特点,采用该算法以一天的交通流量、天气和天的特征(普通工作日、普通假期或者特长假期)为指标,将历史数据分类成若干个类。(3)针对分类完成的每一个类,采用GA-WNN方法建立相应的预测模型,对实际数据进行预测,仿真结果表明:该方法不仅对工作日的交通流量预测精度好于只采用GA-WNN,而且对普通假期、特长假期或者不同天气状况的交通流量预测精度也较好。(4)分析了交通路网的路径选择,基于流量预测设计了路径选择系统,该系统将Dijkstra最短路径算法应用于交通中路径选择为用户提供出行路径方案,并进行了仿真实验,结果表明该系统符合设计需求。
[Abstract]:With the increasing of urban vehicle ownership, vehicle exhaust pollution and road traffic congestion become more and more serious. Traffic guidance system can effectively reduce vehicle pollution and alleviate traffic congestion. Accurate prediction of urban road traffic flow plays a very important role in traffic guidance, and reasonable route optimization can provide a scheme for traffic guidance, thus achieving the purpose of solving traffic problems. In order to improve the accuracy of traffic flow prediction, this paper proposes a neural network method to predict traffic flow data after clustering analysis. The main contents and results of this paper are as follows: (1) analyzing the principle of traffic flow forecasting technology and route selection technology, the research status and application.) aiming at the characteristics and influencing factors of traffic flow data, A traffic flow prediction method based on K-Means algorithm and genetic algorithm (GA) optimization based on wavelet neural network (WNN) is proposed. Because the K-Means algorithm can deal with large scale numerical data quickly and efficiently, the algorithm is based on the characteristics of traffic flow, weather and day (ordinary working day, ordinary holiday or extra long holiday). Classification of historical data into several classes. (3) for each class completed by the classification, the GA-WNN method is used to establish the corresponding prediction model to predict the actual data. The simulation results show that not only the precision of traffic flow prediction is better than that of GA-WNN only, but also the traffic flow prediction accuracy of ordinary vacation, special vacation or different weather conditions is better than that of GA-WNN.) the route selection of traffic network is analyzed. A path selection system based on flow prediction is designed. The Dijkstra shortest path algorithm is applied to provide the route scheme for users in traffic. The simulation results show that the system meets the design requirements.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491

【相似文献】