This study presents a digital twin (DT)-based production analysis and predictive system designed to support real-time production plan validation and decision-making. Unlike traditional simulation models that rely on static historical data, the proposed approach integrates real-time operational data within a DT framework, enhancing predictive accuracy and responsiveness in manufacturing processes. The system framework, operational procedures, and information model were developed to enable continuous monitoring and schedule validation, with its effectiveness demonstrated through a case study in rolling stock manufacturing. The results show that the proposed system improves schedule validation, enhances production forecasting reliability, and mitigates operational uncertainties. Future research will focus on expanding the capabilities of the DT framework to further enhance production flexibility and adaptability, contributing to the development of a more advanced and intelligent manufacturing decision-support system.