汽车未来行驶车速预测

发布时间：2018-06-26 05:19

本文选题：汽车 + 车速预测　；参考：《重庆大学》2015年硕士论文

【摘要】：汽车车速预测的研究已经成为智能汽车以及智能交通领域的研究重点之一。基于车联网预测汽车未来行驶车速,可用于汽车路径导航、碰撞预警及节能辅助驾驶系统,其研究具有重要的理论与应用价值。本文的主要研究内容如下:①通过VISSIM交通流仿真软件构建市区路网模型,模拟实际交通的不同交通状态,包括拥堵、畅通、缓行等交通工况,获得车速预测的原始数据。建立了路段实时动态划分模型,将变长划分与定长划分结合,对汽车未来行驶路段进行细致划分,确定基本路段及其子路段合成的划分方法。②建立两种灰色车速预测模型,即简单灰色预测模型GM(1,1)与灰色神经网络模型GNN。研究表明简单灰色车速预测模型GM(1,1)适用于车辆稳速行驶、趋于加速行驶、趋于减速行驶的车速轨迹预测中,灰色神经网络则适用于拥堵路段走走停停工况。分析了不同时间间隔对车速预测模型预测时长与精度的影响,并验证了两种灰色车速模型的预测精度。③建立最相似近邻车速预测模型,在路段划分的基础上提取路段平均车速,构建当前路段的历史数据库,并从中找到与当前路段平均速度变化特性最为相似的路段历史平均车速序列,作为未来多个时刻路段平均速度的预测值。分析了不同采样间隔与样本时间的历史数据库对最相似近邻预测模型的影响,并验证了模型的预测效果。④通过4种典型工况路段内所有车辆的车速轨迹、加速度及车间距分布数据,分析了7个基本特征参数在畅通、缓行、小拥堵、大拥堵这4种典型工况下的箱线图,确定路段平均速度、速度均方差作为工况识别的2个参数,并制定了工况识别的判断规则。最后论文给出了灰色车速预测模型、最相似近邻车速预测模型、基于其它车辆速度轨迹近似预测方法及路段平均车速近似预测方法的适用条件与范围。
[Abstract]:The research of vehicle speed prediction has become one of the key points in intelligent vehicle and intelligent transportation field. The prediction of future vehicle speed based on vehicle networking can be used in vehicle path navigation, collision warning and energy-saving driving system, which has important theoretical and application value. The main contents of this paper are as follows: 1. The traffic flow simulation software of VISSIM is used to build the urban road network model to simulate different traffic conditions, including congestion, smooth and slow traffic conditions, and to obtain the original data of speed prediction. In this paper, a real-time dynamic partition model of road sections is established, which combines variable length partition with fixed length division, and then makes a detailed division of the future segment of the vehicle, and establishes two kinds of grey speed prediction models, which are composed of the basic road sections and their sub-sections. That is, the simple grey prediction model GM (1 + 1) and the grey neural network model (GNN). The research shows that the simple grey prediction model GM (1 / 1) is suitable for vehicle speed control, tends to speed up, and tends to decelerate, and grey neural network is suitable for stop and stop condition in congested sections. The influence of different time intervals on the prediction time and accuracy of the vehicle speed prediction model is analyzed, and the prediction accuracy of the two grey speed models is verified. 3. The most similar nearest neighbor speed prediction model is established. Based on the division of road sections, the average speed of road sections is extracted, the historical database of current sections is constructed, and the sequence of historical average speed of sections most similar to the changing characteristics of average speed of current sections is found. As the predicted value of the average speed at many times in the future. The influence of historical database with different sampling interval and sample time on the most similar nearest neighbor prediction model is analyzed, and the prediction effect of the model is verified. 4. The speed track of all vehicles in four typical working conditions is verified. The data of acceleration and vehicle spacing distribution are analyzed. The box diagrams of seven basic characteristic parameters are analyzed under the four typical operating conditions: smooth, slow, small congestion and heavy congestion. The average speed of road section is determined, and the mean variance of speed is taken as two parameters to identify the working conditions. The judging rules of working condition identification are also established. Finally, the paper gives the grey speed prediction model, the nearest neighbor speed prediction model, the applicable conditions and the range of the approximate prediction method based on other vehicle velocity trajectory and the road section average speed prediction method.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495

【相似文献】