Retracted: Real-Time Traffic Estimation with Incomplete Information under Urban Traffic Network

It is difficult for managers to do real-time traffic estimation under urban road network because of the incompletion of information collection and the complexity of road network. This paper will improve the precision of data expansion algorithm presented by Lederman R and Wynter L. (2011) using Kalman Filter during real-time estimation. Computation is composed of a heavyweight offline calibration and a lightweight online estimation. During offline calibration, traffic flow is supposed to correspond to user-equilibrium. All historic link flow at a given time window can be derived by solving a mathematical programming. Splitting probability at a specific intersection would then be obtained. Furthermore, all real-time link flow could be estimated preliminarily under the assumption of splitting probability which is supposed to be identical at the same time window and intersection. Afterwards, link flow will be amended by Kalman filter. Simulation is implemented on an urban road network located in XuanWu District, Nanjing. The precision of this algorithm shows that while real-time traffic data lacking of 80% and also influenced by a stochastic disturbance, the average relative error (ARE) is less than 16%. The value of ARE is under 10% while 80% of real-time link value is observed.