Railway point is a critical movable track element in railway infrastructure, its failures mostly cause risks of derailment and collision. Railway point machine is used for the switching, locking, and supervising railway point, its maintenance is crucial to guarantee the availability of railway point. With the wide deployment of sensors and the rapid development of digital technology, the pure detection methods cannot meet the current demand for the fault detection and diagnosis of railway point machines. As a way to build a closed loop between the physical object and the virtual model, the digital twin can realize the identification of possible root causes of malfunctions. In this paper, we proposed an extended framework for digital-twin-assisted fault diagnosis of RPM, which supports the realization of the accurate and real-time monitoring and diagnosis of the fault of the railway point machine. This proposed framework is developed based on the three-dimension and five-dimension structure of the digital twin model. In the case study, the predefined failures and a method based on the electric current value and shape for digital-twin-assisted fault diagnosis of RPM are presented.