Maintenance of train doors is an important aspect of railway upkeep, which is critical to ensure the safety and punctuality of railway operations. In order to assist frontline maintenance staff in efficiently and effectively carrying out the required maintenance operations, while minimizing downtime associated with train door maintenance, this paper proposes a method that integrates digital twinning and augmented reality (AR). The method consists of two main parts: (1) Based on digital twinning technology, a virtual model of the train door, integrated with real-time sensor data, provides a live representation of door operational status. (2) Using sensor data and maintenance records, a fault tree for train doors was constructed, enabling failure diagnosis. (3) Using AR, frontline staff are provided with immersive assistance during their maintenance activities. To validate the effectiveness of this method, an AR prototype deployed on Microsoft Hololens2 is developed. The prototype uses digital twin data to help maintenance staff quickly locate the component that is faulty and provides step-by-step maintenance guidance, thereby optimizing the maintenance processes.