基于机器学习的地铁行人流在线优化控制研究  

Machine Learning-based Simulation Research of On-line Subway Pedestrian Flow Control

作　　者：史佳洁 杨鹏[1] 皮雁南 Shi Jiajie;Yang Peng;Pi Yannan(Tianjin University of Technology,Tianjin 300000,China;Metro Operation Technology R&D Center,Beijing Metro Operation Co.,Ltd.,Beijing 100082,China)

机构地区：[1]天津理工大学,天津300000 [2]北京市地铁运营有限公司地铁运营技术研发中心,北京100082

出　　处：《系统仿真学报》2023年第2期386-395,共10页Journal of System Simulation

基　　金：中央高校基本科研业务费(2019JBM032)。

摘　　要：为了实现高峰期地铁站行人流管控的在线优化,设计了基于机器学习的地铁站行人流管控算法框架。以某地铁车站早高峰的行人流管控流程为研究对象,利用Agent技术搭建地铁站行人流管控仿真模型。多次运行仿真模型可以获得深度学习网络的训练数据。通过对网络进行充分训练,得到优化调度策略。将网络接入地铁站行人流实时运行数据,实现实时优化管控。仿真实验表明:引入的深度强化学习框架可以实现在线优化,调度结果优于传统方法。For the online optimization of pedestrian flow control in subway station, an algorithm frame for pedestrian flow control in subway station based on machine learning is designed. The pedestrian flow control process of a subway station during morning rush hour is selected, and the agent-based model is built to simulate the control process. The training data is collected through the multiple runs of the model,which is used as the input of deep reinforcement learning network, and the mature net is obtained through adequate training to provide the optimizing scheduling policy. Linking the actual data with the mature net to realize the real-time schedule optimization of subway pedestrian flow control. Simulation experiments show that the framework of the deep reinforcement learning can realize the on-line optimization and the performance is better than traditional algorithm.

关 键 词：深度强化学习 行人流管控 地铁站调度策略 在线仿真 实时优化 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]