Lane-based Optimization for Coordinated Traffic Signal and Network Configuration Designs
DescriptionThe lane-based mathematical programming approach was recently developed by the investigator for designs of individual and linked signal-controlled junctions. Conventionally, traffic flows and geometric aspects are given as exogenous inputs to design signal settings. The present study will combine the geometrical junction layout and signal controls for designing equilibrium signal-controlled network. Feasible vehicle paths/routeings connecting different origin and destination locations in an urban network will depend on the junction geometric aspects, connectivity of consecutive junctions, traffic lane usages, and permitted (or ban) turns on approach lanes at junctions. Once the lane usages and permitted turns on approach traffic lanes are designed, lane markings will be painted on ground showing road users the permitted turn directions on traffic lanes as guidance for making turns at junctions within an urban network. Actual performances of an urban network depend on total delay experienced by all users and lengths of vehicle queues. Good junction configurations are fundamentally important for network link connections. Therefore, there is a strong need to integrate the geometric aspects and signal control aspects together for designing a better network system. The design problem is a complex one and mathematically non-linear and non-convex in nature that can be formulated as a three-stage optimization problem as proposed. Network configuration will be defined by the feasible turn directions from upstream junctions to the downstream junctions at the first-stage. At the second-stage, individual lane usages, lane marking patterns and aspects of signal settings represented by the binary-type variables at individual junctions will be determined. These can generate a computer network for detailed modeling and evaluations. The third-stage will be a group-based optimization of signal timings including offsets, cycle time, and other signal timings. Also, a path-based assignment for traffic equilibrium network with special considerations of actual vehicle queue development will be solved. It prevents overflowing and blocking back to upstream junctions due to ineffective signal settings. The three-stage problem is solved by a proposed heuristic procedure that involves a genetic algorithm, a standard branch-and- bound technique, and a Newton’s gradient-based optimization algorithm.Study findings will be able to enhance the practical design of signal-controlled networks in urban cities where junctions are closely spaced. Better geometric designs for lane usages and turn directions on approach lanes, better traffic signal control settings with proper sequences and precise signal timings, and overall less delay and shorter vehicle queues for users traveling in a signal-controlled network will be achieved.
|Effective start/end date||1/01/14 → 13/12/17|