摘要
为了提出一种适用于任何流量并具有较高检测率和较低误警率的快速路交通事件检测算法,以统计理论、突变理论为基础,设计了事件影响指数检测算法.利用交通量数据,检验纵向时间序列与横向时间序列上流量的波动性和正态拟合性;分析事件数据与非事件数据的差异,得出交通事件数据变化特征.分析结果表明:纵向时间序列的波动性和正态拟合性优于横向时间序列;事件数据具有多模态、不可达、突跳等突变性特征.该算法误警率为0,检测率比经典的Cali-fornia算法高出10%.不同流量下的检测效果对比表明,该算法适用于各种流量,低流量状态下的检测效果更好.
In order to put forward a novel traffic incident detection algorithm for expressway which can apply to any traffic volume with a high detection rate and a low false alarm rate,an incident influence index detection algorithm is designed based on the statistical theory and the catastrophe theory.The stability and normal goodness-of-fit of longitudinal time series and transverse time series are tested using traffic volume data.The characteristics of traffic incident data are obtained by analyzing the difference between incident data and non-incident data.Analysis results show that the stability and normal goodness-of-fit of longitudinal time series are much better than those of transverse time series,and traffic incident data exhibits multi-modal,unreachable,and sudden jump characteristics.Compared with the California algorithm,the false alarm rate of the proposed algorithm is 0,and the detection rate is 10% higher than that of the California algorithm.The comparison of detection results under different volumes shows that the proposed algorithm is suitable for any volumes,and the detection effect under low volume is even better.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2013年第3期649-653,共5页
Journal of Southeast University：Natural Science Edition
基金
国家自然科学基金资助项目(51078232)
关键词
交通事件
快速路
环形线圈
纵向时间序列
突变理论
事件检测
traffic incident
expressway
loop coil
longitudinal time series
catastrophe theory
incident detection
