基于改进BP神经网络的交通流数据融合

Traffic Flow Data Fusion Based on a Modified BP Neural Network

提出一种改进BP神经网络的交通流数据融合算法,通过对融合模型以及融合算法的改进,实现融合精度及容错能力的提高,最终利用某城市主干路的交通流实测数据进行实验验证.验证结果表明,本文提出模型可以通过LSE值验证且融合精度可以达到94%以上,因此利用改进BP神经网络可以有效进行交通流参数的融合,并可以较为准确反映路段平均速度的变化,为交通流的预测与诱导提供理论支持.

A traffic flow data fusion algorithm is proposed based on an improved BP neural network in this paper. The model proposed to enhance the fusion algorithm so that the fusion accuracy and fault-tolerant ability can be improved significantly. The data is used to verify the precision of the fusion experimq =kuent,and collected by the detection in one main road in a Mega city. Verification results show that the use of the BP neural network can effectively fusion the traffic flow parameters,whose fusion accuracy can be greater than 94%. The results more accurately reflect the change of the average road speed in the proposed model. It can provide the theoretical support for the prediction of traffic flow and the induction.It can also provide the basis for scientific,intelligent traffic organization and management,and intelligent control.