Post-disaster Highway Network Restoration Decision Based on Reinforcement Learning

被引：0
|
作者：
Hao X.-J. [1 ]
Mao X.-H. [2 ,3 ,4 ]
Tan X.-Y. [2 ,3 ,4 ]
Wang J.-W. [2 ,3 ,4 ]
机构：
[1] School of Management, Xi'an University of Finance and Economics, Shaanxi, Xi’an
[2] College of Transportation Engineering, Chang’an University, Shaanxi, Xi’an
[3] Engineering Research Center of Digital Construction and Management for Transportation Infrastructure of Shaanxi Province, Chang’an University, Shaanxi, Xi’an
[4] Xi’an Key Laboratory of Digitalization of Transportation Infrastructure Construction and Management, Chang’an University, Shaanxi, Xi’an
来源：
Zhongguo Gonglu Xuebao/China Journal of Highway and Transport | 2023年 / 36卷 / 08期
基金：
中国国家自然科学基金;
关键词：
highway network restoration decision; reinforcement learning; resilience; traffic engineering;
D O I：
10.19721/j.cnki.1001-7372.2023.08.026
中图分类号：
学科分类号：
摘要：
Efficient decision-making in the aftermath of disasters is pivotal for rapidly restoring highway network connectivity, ensuring seamless disaster relief operations such as pedestrian evacuation, emergency learn dispatches, and the transportation of disaster relief materials. This paper delves into the scheduling and routing of highway network repair crews during emergency recovery phases. Il emphasizes flexible scheduling, allowing multiple repair crews lo concurrently address a single damaged road segment, enhancing the strategy's real-world applicability. Connectivity of the post-disaster road network is quantified using the demand satisfaction rate of affected points. This study introduces "road network connectivity resilience" as a metric lo assess the resilience during the repair process. A Markov decision process is employed to simulate the decision-making involved in repair crew scheduling and routing. To address the decision-making challenge, an enhanced algorithm, amalgamating Q-learning and Dijkstra algorithm, is introduced. A comprehensive case study reinforces the efficacy of the proposed method. Results reveal that this approach facilitates a holistic decision-making process for repair crew scheduling and routing, yielding an optimal repair strategy with superior global resilience. The findings further underscore that factoring in flexible resource scheduling leads to a more refined and pragmatic repair strategy. © 2023 Xi'an Highway University. All rights reserved.
引用
收藏
页码：292 / 304
页数：12
相关论文
共 26 条
  • [21] ZHANG Guo-fu, CHANG Jia-yuan, SUZhao -pin, Et al., Repair crew scheduling for damaged road network with enormous demand points using deep Q-learning [J], Control and Decision, 37, 12, pp. 3267-3277, (2022)
  • [22] SU Zhao-pin, LI Mo-han, ZHANG Guo-fu, Et al., Modeling and solving the repair crew scheduling for the damaged road networks based on Q-learning [J], Acta Automatica Sinica, 46, 7, pp. 1467-1478, (2020)
  • [23] WANG W, FAN J, PENG X, Et al., Maintenance worker scheduling for charging pile fault: A multi-agent RL approach, 2022 IEEE International Conference on Energy Internet (ICEI), pp. 82-89, (2022)
  • [24] MAO X H, ZHOU J B, YUAN C W, Et al., Resilience-based optimization of postdisaster restoration strategy for road networks, Journal of Advanced Transportation, 2021, pp. 1-15, (2021)
  • [25] KIM S, SHIN Y, LEE G M, Et al., Network repair crew scheduling for short-term disasters, Applied Mathematical Modelling, 64, pp. 510-523, (2018)
  • [26] TIAN Qi-dong, ZHANG Jia-qi, CHEN Ying, Et al., Postdisaster emergency repair and reconfiguration plan of faulty distribution system based on reinforcement learning [J], Advanced Technology of Electrical Engineering and Energy, 42, 3, pp. 66-75, (2023)
123