A multi-objective integrated large-scale optimized ramp metering control system for freeway/surface-street traffic management

Abstract

This research, denoted MILOS (Multi-objective Integrated Large-scale Optimized ramp metering System) is a hierarchical structure for solution of the large-scale freeway management problem to address the key features of this problem (dynamic state changes, stochasticity, multi-dimensionality, unpredictability, partial-observability, and existence of multiple objectives). MILOS decomposes the freeway control problem into subproblems along temporal/spatial boundaries and is composed of three primary components: SPC-based anomaly detection and optimization scheduling, area-wide coordination layer, and predictive-cooperative real-time (PC-RT) optimization layer. The area-wide coordination component of the hierarchical control system considers the impact of queue growth on the adjacent interchanges in a quadratic programming optimization model with a multi-criterion objective function. The formulation of the area-wide optimization problem is augmented with overflow variables to guarantee a feasible solution. The nominal solution of the areawide coordination problem is then modified in real-time by the locally traffic-reactive, PC-RT algorithm based on a linear-program using a linearized dynamic difference equation implementation of the macroscopic FREFLO model. The PC-RT formulation pro-actively plans to utilize opportunities to disperse queues or hold back additional vehicles when freeway and ramp demand conditions are appropriate. The cost coefficients of this optimization problem is linked to the solution of the area-wide coordination problem by using information on the dual of the solution to the area-wide coordination problem. The optimization runs of the area-wide coordination problem and the PC-RT optimization problems at each ramp are scheduled by a demand/flow monitoring system based on statistical process control. A simulation experiment is executed to evaluate the MILOS hierarchical system against "no control", ADOT's current ramp metering policy, and an area-wide LP optimization problem resolved in 5-minute intervals on a small freeway network in the metropolitan Phoenix, AZ area. Three test cases are presented for a short "burst" of heavy-volume flows to all ramps, a 3-hour commuting peak, and a 3-hour commuting peak with a 30-minute incident occurring in the middle of the network. The performance results indicate that MILOS is able to reduce freeway travel time, increase freeway average speed, and improve recovery performance of the system when flow conditions become congested.