|
|
|
| Bi-level Programming for Resilience Restoration of Commuting Corridor Based on Deep Reinforcement Learning |
| LI Xueyana, ZHANG Tongyub, ZHU Xina
|
| a. School of Management; b. School of Urban Rail Transit and Logistics, Beijing Union University, Beijing 100101, China |
|
|
|
|
Abstract In order to realize the scientific design of motor bus transferring scheme in the commuter corridor, the resilience recovery process of commuting corridor is regarded as a bi-level programming in which the resilience is improved through continuous exploration and iteration of ground bus transferring scheme in complex environment. The deep reinforcement learning algorithm is introduced to form the upper level planning, and the value function neural network is used to fit the response function of emergencies and travelers' cluster behavior to the adjustment of ground bus transferring scheme. The decision-making objective is achieved by training the transferring schemes. In the lower level planning, the cellular neural network model is introduced to simulate the cluster travel choice behavior under the background of data intelligence. The case study shows that this method can effectively improve the resilience of the commuter corridor, and the cluster behavior will have a negative impact on the resilience recovery.
|
|
Received: 10 May 2022
Published: 26 April 2024
|
|
|
|
|
|
|
|
|
