Traffic control optimization strategy based on license plate recognition data

Traffic signal control is essential to the efficiency of the road network's operation. In recent years, more and more detailed detection data provide potential data support for traffic signal control, such as license plate recognition (LPR) data. This study aims to develop a traffic signal control optimization method based on model predictive control (MPC) and LPR data. The proposed framework of a closed-loop control system is described in detail. First, the control objectives and queue prediction model for signalized intersection are determined. Then, online optimization and feedback compensation are discussed and implemented. Calculations of the arrival rate at the downstream are based on the LPR data detected at the upstream intersection, and dynamic optimization method of the offset is proposed for a coordinated control. The model is validated using the LPR data of two consecutive intersections with a traffic simulation platform. Results demonstrate that the model can restrain extreme long queuing, improve intersection capacity, and reduce intersection average delay. The developed model promotes the system operating efficiency and shows the general advantage of real-time optimization, feedback, and control. The proposed framework can be potentially applied by local traffic management centers to improve the quality of traffic signal control. •Implemented online optimization and feedback compensation in the control system.•Applicable to traffic control system with license plate recognition data.•Validated the framework with close-loop experiments in VISSIM and MATLAB.•Demonstrated improvements in queue and delay reduction for coordinated intersections.