Predictive Ramp Metering Control Based on Physics-Informed LSTM with Queue Length Constraints on Entering Ramps

Adaptive ramp metering techniques are necessary for efficient freeway traffic management in order to avoid congestion and reduce delays. In order to improve traffic flow and control excessive queue building at on-ramps, this article proposes a Queue Override Mechanism into the AI-based Predictive ALINEA controller. In order to ensure effective on-ramp utilization and avoid queue spillback, the suggested improvements dynamically modify the metering rate based on queue length and anticipated downstream traffic conditions. Using key performance indicators, the performance of the Enhanced Predictive ALINEA controller is assessed in comparison to regular ALINEA, Predictive ALINEA, and actual system behavior. Tested on a real case study, the suggested strategy maintains effective freeway flow while drastically cutting onramp wait times.