Smart Operation Platform for Railway Rolling Stock Production Using Digital Twin and AI

Digital twin-based production operation platforms play a critical role in managing complex process flows and workflows in large-scale assembly manufacturing environments. They enable stable operational decision-making under unexpected disruptions. Most existing studies on digital twin-based platforms, however, focus on monitoring or post hoc validation. This study proposes a digital twin-based intelligent production operation platform that integrates AI-enabled real-time shop-floor monitoring, discrete-event simulation, and AI-driven scheduling optimization within a closed-loop decision-making framework. The proposed platform diagnoses the operational impacts of disruptions in advance, regenerates alternative schedules, and validates executability through simulation prior to deployment. Additionally, when unexpected abnormal situations occur, the platform supports stable shop-floor operation by minimizing delay propagation through the application of validated alternative schedules. Its effectiveness is demonstrated through a case study in a large-scale assembly industry. The results show that the proposed platform effectively manages delay propagation and stabilizes workload distribution while significantly improving throughput and due-date performance, thereby enhancing adaptability and operational efficiency in complex manufacturing environments.