Loop detector data error diagnosing and interpolating with probe vehicle data

Traffic volume and queue length are two important variables to identify traffic states in urban areas. Loop detectors are often used for monitoring traffic. A prominent weakness of loop detector data is the limited reliability due to equipment malfunctions and communication faults. This paper presents a method to identify the validity of loop detector measurements by analyzing the ratio of counts on adjacent lanes. GPS data from probe vehicles are an alternative for monitoring traffic states. They include instantaneous speed, acceleration and position; individual vehicle trajectories can be grossly estimated in real time. Consequently, the queue length can be measured approximately. This can be used to estimate traffic volumes and the percentage of probe vehicles. By combining loop detector and GPS data, more information regarding traffic status can be extracted from both data sources. This paper discusses two methods to check the volume counts and to fill in missing or invalid data: one method uses the ratio of loop detector counts on adjacent links, the other one uses data from probe vehicles to estimate the traffic volumes from the dynamics of the queue length at signalized intersections. Real data from the city Changsha in China are used to validate these two methods. The developed methods provide a way for the online monitoring of detectors’ performance and level of service at signalized intersection