No open-access location is known for this record yet.
No open-access location is known for this record yet.
Simultaneous optimization of network traffic dynamics and travel behavioral responses for decision makers under increasingly popular advanced traveler information systems and advanced traffic management systems requires the use of appropriate tools, which capture the joint impact of various strategies on the transportation system. Developing analytical models incorporating both traffic dynamics and travel behavior is challenging. Although emerging tools integrating agent/activity based models and dynamic traffic assignment have been developed, these tools are typically used for the performance evaluation instead of optimization because of the expensive development and implementation in large-scale networks. To get rid of these limitations, this paper proposes a simulation-based optimization approach, i.e. an innovative Bayesian stochastic Kriging metamodel, which adapts for the heteroscedasticity of simulation noises to solve the simultaneous optimization problem of travel behavioral responses and dynamic traffic management. Simulation-based optimization combines the advantages of both simulation models and mathematical optimization methods and only requires a limited number of objective function evaluations. Transportation system simulation outputs are heteroscedastic because of a variety of probabilistic inputs, e.g. random seed, signal timing, route choice behaviors. The proposed metamodel takes into account the model uncertainty in a Bayesian framework. In a work zone scenario on a real-world freeway/arterial corridor of I-270 and MD 355 in the State of Maryland, the proposed Bayesian stochastic Kriging model is applied to the simultaneous optimization of the high-occupancy/toll rate and freeway diversion rates via congestion warning at two upstream off-ramps. The objective is primarily to improve the corridor-wide performance. Field measurements by loop/microwave detections are used to calibrate travel demand and supply models. Results show that the simultaneous optimization is promising in reducing the corridor-wide travel delay and in enhancing vehicle throughputs. The approach presented in this paper is general and can be extended to simultaneously optimize many other efficient real-world control strategies for large-scale networks.
Loading…
Imported from OpenAlex (CC0) — nothing here is generated.