Estimation of Queuing Length at Signalized Intersections Using Low-frequency Point Detector Data

Due to the limitation of traffic data collected at signalized intersections in most small and medium sized cities in China,estimation on queuing length at signalized intersections is with low accuracy.In order to solve this prob-lem,a model of estimating queue length based on data from low frequency,fixed-point,and single detectors at signalized intersections is proposed.A long queuing event(the end of a queue exceeds the position of the detector)is identified based on the relationship between occupancy,flow,and speed.Data of detectors is divided into cycles according to and synchro-nized with signal timing.The maximum length of queuing is calculated based on shock wave theory.A simulation test and a demonstration study are carried out at the south entrance lanes of Shandong road and Jiangxi road in Qingdao.The re-sult shows that identification accuracy of a long queue is above 90%.The average estimated accuracies of queuing length under different saturation(low,medium,and high)are all more than 80%,among which,the average MAE is lower than 20 m per cycle under medium and low saturation,and lower than 45 m per cycle under high saturation.