摘要
面对大量的交通参数数据,如何快速建立高效的分类预测模型以尽快地对拥堵状态进行判别是一个重要的问题。本文利用检测器提供的车速、流量和占有率等相关参数信息,提出了基于学习向量量化(LVQ)神经网络的交通拥堵预测模型。通过使用Matlab7.3数学软件的神经网络工具箱对该神经网络不断地训练,最终可以对道路正常状态和拥堵状态进行分类,通过交通流参数数据的输入预测出路段交通拥堵状况。最后,给出算例进行网络训练和测试,训练结果表明,运用该算法进行交通拥堵预测取得了良好的效果,具有一定的现实意义。
For a large number of traffic parameter data, it is an important issue that how to set up an efficient model of classification and prediction to identify the congestion state as soon as possible. The article provided amodel to predict traffic congestion based on the learning vector quantization neural network by using the traffic parameters such as speed, traffic flow, road occupancy, etc. ,which were detected by vehicle detectors The model can finally classify the traffic congestion and normal situation by continuously training the LVQ neural network with the software matlab. The model can predict the road traffic situation by inputing the traffic flow data. Finally, an example was given to train and test the network. The training result demonstrated that the algorithm was feasible to predict traffic congestion and is actually useful in reality.
出处
《交通运输工程与信息学报》
2009年第3期97-101,106,共6页
Journal of Transportation Engineering and Information
基金
"十一五"国家科技支撑计划(2006BAJ18B01)
国家高技术研究发展计划(2008AA11Z201)
