Real-time Decision Support System for Transportation Infrastructure Management Under a Hurricane Event
-
2022-08-01
Details:
-
Creators:
-
Corporate Creators:
-
Contributors:
-
Corporate Contributors:Rutgers University. Center for Advanced Infrastructure and Transportation ; United States. Department of Transportation. University Transportation Centers (UTC) Program ; New York State Department of Transportation ; United States. Department of Transportation. Office of the Assistant Secretary for Research and Technology ; United States. Department of Transportation. Federal Highway Administration ; ... More +
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report Feb. 1, 2021 – Jan. 31, 2022
-
Corporate Publisher:
-
Abstract:Under hurricane weather and traffic conditions, stakeholders need to make a series of decisions to close or restrict the traffic of vulnerable components in the transportation network for balancing traffic safety and mobility. To manage these critical components for minimizing the overall network-level losses, a deep reinforcement learning (RL)-based decision support system is employed. Specifically, the stochastic sequential decision problem of managing hurricane-impacted infrastructures is formulated as a Markov decision process, which is solved by RL methodology with deep neural network-based function approximations for the traffic control policy. It is noted that the deep RL-based minimization of overall network-level losses essentially sacrifices the traffic safety (in terms of vehicle accident risk) to obtain a significant benefit from traffic mobility (in terms of travel time), which may be unacceptable for certain risk-averse stakeholders. To address this issue, intelligent travel advisories broadcasted through various media channels are utilized, as an additional action in the RL framework, to actively redistribute the travel demand to time periods with relatively low hurricane intensity. Accordingly, the overall network-level cost can be mitigated without greatly increasing the traffic-safety losses. For concept proof, a case study on a hypothetical transportation network under hurricane events is used to demonstrate the good performance of the newly developed deep RL-based decision support system.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
File Type:
Supporting Files
-
No Additional Files
More +