融合多源数据的ABC-SVM快速路交通事件检测  

Urban expressway's traffic incident detection based on multi-source data fused by ABC-SVM

作　　者：丁宏飞[1,2] 秦政[2] 李演洪 刘博[2] 

机构地区：[1]四川省交通运输厅公路规划勘察设计研究院,四川成都610041 [2]西南交通大学交通运输与物流学院,四川成都610031

出　　处：《中国安全科学学报》2015年第6期162-166,共5页China Safety Science Journal

基　　金：国家自然科学基金资助(51308475);四川省科技支撑计划资助项目(2011FZ0050)

摘　　要：为保证城市快速路段的畅通,建立一种基于蜂群算法-支持向量机(ABC-SVM)融合多源交通数据的城市快速路交通事件检测方法。首先通过分析路段实际情况,选取不同检测器的数据作为事件检测模型的输入值;然后利用蜂群算法(ABC)对支持向量机(SVM)分类模型中的参数进行优化,获得最优的交通事件检测模型,模型的输出结果为检测路段是否发生交通事件;最后结合成都市三环城市快速路路段上采集到的多源交通数据进行实例验证。结果表明,利用ABC-SVM方法进行事件检测的效果优于BP神经网络的方法。For guaranteeing urban expressway unblocked,the author worked out an urban expressway＇s traffic incident detection method based on multi-source traffic data fused by ABC-SVM. First,different detector＇s field data were analyzed and served as traffic incident detection model＇s input. Then,ABC algorithm were used for SVM classification model＇s parameter calibration,thereby an optimized traffic incident model was acquired,of which the output is a Boolean value indicating whether or not incidents occurred in detection links. Finally,a case study based on Chengdu ＇s 3rdring road ＇s multi-source traffic data was carried out. The result shows that ABC-SVM incident detection method performs better than BP neural network method.

关 键 词：城市快速路 交通事件检测 多源交通数据 蜂群算法(ABC) 支持向量机(SVM) 

分 类 号：X924.2[环境科学与工程—安全科学]