A traffic elastic control method and system based on reinforcement learning

Abstract

The invention relates to a traffic elastic control method and system based on reinforcement learning. The method includes: dividing the transportation network into several regions, and collecting traffic flow information in each region; determining the macro basic map of each region according to the traffic flow information; Determine the imbalance rate of the traffic network horizontally; according to the imbalance rate, determine the critical imbalance threshold of the traffic network through seepage analysis; establish a multi-agent reinforcement learning model, and perform multi-agent reinforcement learning model according to the traffic flow information, flow level and critical imbalance threshold. Learning and training, generating a trained multi-agent reinforcement learning model; using the trained multi-agent reinforcement learning model to regulate the actual traffic network, so that the current imbalance rate of the actual traffic network is less than the critical imbalance threshold. The invention can achieve the best regulation effect under the shortest regulation cycle.

Description

technical field [0001] The invention relates to the interdisciplinary field of traffic elasticity, traffic control and network science, in particular to a traffic elasticity control method and system based on reinforcement learning. Background technique [0002] In recent years, catastrophic emergencies still occur from time to time, and these events often cause huge social and economic losses. For example, the Suez Canal ship blockage incident that occurred in March 2021 was caused by a cargo ship encountering a cross wind while passing through the Suez Canal and the hull was swayed, directly blocking the canal channel. According to relevant estimates, due to the impact of this incident, the Suez Canal may lose about US$400 million for every day of blockage. In view of the possible disastrous consequences of emergencies, how to improve the emergency response capabilities of the urban system is a pain point for urban managers. In this context, Holling proposed the concept ...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a traffic elastic control method and system based on reinforcement learning to solve the problem of long control cycle and poor control effect

Images