摘要
为了提高预测精度,利用每周同一天交通流变化相似的特点,提出了一种短时交通流组合预测模型,采集每周同一天的交通流数据进行预测。组合模型包括两个子模型:BP神经网络模型、GM(1,1)模型。BP神经网络模型具有强大的非线性逼近能力,对于庞大无序的交通流数据信息具有良好的处理能力。GM(1,1)模型能够反映交通流时间序列的总体变化趋势,相对误差小。通过计算两种子模型在上一时间段的预测误差比值,确定出在下一时间段的预测中两种子模型预测结果所占的权重,然后将这两个子模型在下一时间段的预测结果进行加权求和,作为组合模型的最终预测值。实验结果表明,组合模型发挥了两种子模型各自的阶段性预测优势,是短时交通流预测的一种有效方法。
A short -term traffic flow prediction based on combined model is presented in this paper. In order to improve the prediction accuracy, the traffic flow data of the same days in different weeks are used. The combined model has two sub - models ： BP neural network model and GM （ 1,1 ） model. BP neural network model has strong dynamic nonlinear mapping ability and has good processing ability to the huge and disorder traffic flow data informa- tion. GM （ 1, 1 ） model can reflect the overall trend of traffic flow time series and has small relative error. In this pa- per, by calculating the prediction error ratio of two sub - models in the previous time period, the prediction weights of the two sub - models in the next predict are obtained, and then the two individual forecast results are combined by u- sing the obtained weights to get the final forecasting of the traffic flow. The experimental results show that the pro- posed combined model takes the advantage of the BP neural network model and the GM （ 1, 1 ） model and produces more precise forecasting than those by two sub - models. It is an efficient method to the short - term traffic flow fore- casting.
出处
《计算机仿真》
CSCD
北大核心
2015年第2期175-178,193,共5页
Computer Simulation
关键词
智能交通系统
神经网络模型
组合模型
Intelligent traffic system
Neural network model
Combined models
