考虑多特征的高速公路交通流预测模型

Highway Traffic Flow Prediction Model with Multi-features

为准确预测高速公路交通流,缓解高速公路交通拥堵现象,本文提出一种考虑多特征的高速公路交通流预测模型。首先将高速公路当前道路与上下游的交通流、天气等数据转化为一个二维矩阵,并利用滑动窗口模型获得输入样本的最佳长度;然后将样本数据输入集成深度学习模型训练并提取交通流数据的特征,随后输出预测结果;最后,将某高速公路交通流数据用于工作日和节假日两组实验。结果表明:集成深度学习模型比单一模型预测高速公路交通流的效果要好,工作日的高速公路交通流预测精度远高于节假日,本文模型将平均绝对误差由6.40辆·(20 min)^(-1)降到5.450辆·(20 min)^(-1),说明考虑多种因素可以提升高速公路交通流预测精度。

In order to accurately predict highway traffic flow and thus alleviate traffic congestion of highway,a highway traffic flow prediction model with multi-features is proposed in this paper.Firstly,the traffic flow of the section with the upstream and downstream sections,and the weather data are transformed into a two-dimensional matrix,and the sliding window model is employed to obtain the optimal size of input samples.Then these samples are input into a hybrid depth deep learning model to extract the features of traffic flow data,and then output the prediction results.Finally,the traffic flow data of a real highway is used to do two experiments on weekdays and holidays.The results indicate that the hybrid deep learning models perform better results than single models for forecasting highway traffic flow.The prediction accuracy of highway traffic flow on weekdays is higher than that on holidays.The proposed model reduces the mean absolute error from 6.40 Cars·(20 min)^(-1) to 5.450 Cars·(20 min)^(-1),which shows that the prediction accuracy of highway traffic flow can be improved by considering multiple related factors.