基于AF-SVR的城市快速路多源交通信息融合研究

发布时间：2018-03-29 23:30

本文选题：多源交通信息　切入点：信息融合　出处：《计算机工程与应用》2017年05期

【摘要】：针对单一检测器所得到的交通数据不能够全面准确地反映实际的交通状态,提出一种基于AF-SVR模型的城市快速路多源交通信息融合的方法。首先通过将相同路段中不同检测器的速度数据作为学习样本输入到支持向量机回归模型(Support Vector Regression,SVR)中进行训练。然后利用鱼群算法(Artificial Fish,AF)对支持向量机回归模型中的参数进行优化,获得最优的信息融合模型,用于多源交通信息的融合,输出为能准确反映真实交通状态的速度数据,并用人工采集的速度数据作为真值进行验证。最后将此方法应用于成都市三环快速路路段上的多源交通信息融合,取得了令人满意的结果。
[Abstract]:In view of the fact that the traffic data obtained by a single detector can not reflect the actual traffic state comprehensively and accurately, A method of multi-source traffic information fusion for urban expressway based on AF-SVR model is proposed. Firstly, the speed data of different detectors in the same section are input into support Vector regression model (SVM) as learning samples. Then the parameters in the regression model of support vector machine are optimized by using the fish swarm algorithm named artificial Fishery (AFF). The optimal information fusion model is obtained for multi-source traffic information fusion. The output is the speed data which can accurately reflect the real traffic state. Finally, the method is applied to the multi-source traffic information fusion on Chengdu Sanhuan Expressway, and satisfactory results are obtained.
【作者单位】：西南交通大学交通运输与物流学院;四川省交通运输厅公路规划勘察设计研究院;
【基金】：国家自然科学基金(No.51308475) 四川省科技支撑计划资助项目(No.2011FZ0050)
【分类号】：U491;TP18

【相似文献】