Due to excessive traffic demand and regional demand deviations inside the city,

many large cities are suffering from problems such as traffic congestion, air pollution, and lack of parking space. To solve this problem, traffic demand management such as reorganization of road space, inducing modal shift to public transportation, and parking management is being implemented. But more active managements need to be implemented to effectively achieve the goals of traffic management policies. Recently, the effectiveness of active traffic demand management, a method of real-time management of traffic systems and demand management through continuous traffic condition monitoring and traffic state prediction methodology by giving real-time property to existing traffic demand management, has been proven and applied to urban networks. In this study, the effectiveness of active traffic management techniques was analyzed for Hanyang City Wall with entry and exit traffic data. A link travel speed prediction model according to the entry traffic volume was constructed using DA-RNN, and the contribution of each entry point was identified by applying the SHAP algorithm. After that, a management technique using reinforcement learning was developed. The effects of implementation for each scenario were analyzed, and it was found that link travel speed increased by up to 7% depending on the time period. It is expected that more effective analysis will be possible when a digital twin transportation system is built in the future.