高速公路短时交通流量预测方法设计与实现

Design and Implementation of Highway Short-term Traffic Flow Prediction Method

高速公路交通流量受到多个因素的影响,模式复杂多变,难以准确获取短时交通流动特征。为此,引入动态时空神经网络对高速公路短时交通流量进行预测。利用交通流量阈值,计算超出道路通行能力的车辆,根据数据样本之间的欧氏距离,修复交通流量数据。构建基于动态时空神经网络的高速公路短时交通流量预测模型,采用多层三维卷积捕捉短时交通流动特性,获取高速公路短时交通流量预测结果。实验结果表明,所提方法能够较好地拟合真实的交通情况,确定系数高达0.94,预测延误最高仅为0.009 ms。

The highway traffic flow is affected by many factors,the change pattern is complex and changeable,and it is difficult to accurately obtain the short-term traffic flow characteristics.Therefore,a dynamic spatiotemporal neural network is introduced to predict the highway short-term traffic flow.The traffic flow threshold is used to calculate the vehicles exceeding the road capacity,and the traffic flow data are repaired according to the Euclidean distance among data samples.The highway short-term traffic flow prediction model based on dynamic spatiotemporal neural network is constructed.Multi-layer three-dimensional convolution is used to capture the characteristics of short-term traffic flow,and the highway short-term traffic flow prediction result is obtained.The experimental results show that the proposed method fits the real traffic conditions well,the determination coefficient is as high as 0.94,and the maximum predicted delay is only 0.009 ms.