基于交通小区短时交通生成预测的城市道路网短时动态交通预测研究

Research on Short-time Dynamic Traffic Prediction of Urban Road Network Based on Short-time Traffic Generation Prediction of Traffic Zones

城市道路周边用地性质的不同导致短时发生和吸引交通量的差异,直接影响城市道路网上的交通流。论文对城市路网中各地块的发生和吸引交通量进行短时预测,将预测的交通量通过动态交通分配到城市路网中,以此实现城市路网短时动态交通预测。将单一的BP神经网络模型为对比模型,对模型进行训练及参数标定,并检验模型的预测效果。实验结果表明,论文所提的预测方法与BP神经网络相比,预测精度最高可提升84.28%。

The difference of land use nature around urban roads leads to the difference of short-time occurrence and attraction traffic volume,which directly affects the traffic flow on the urban road network.This paper makes short-time prediction of occur-rence and attraction traffic volume of each plot in the urban road network,and allocates the predicted traffic volume to the urban road network through dynamic traffic,so as to realize short-time dynamic traffic prediction of the urban road network.The single BP neural network model is used as the comparison model,and the model is trained and parameter calibrated,and the prediction effect of the model is tested.The experimental results show that the prediction accuracy of the proposed prediction method can be improved by up to 84.28%compared with that of BP neural network.