Solid Rocket Motor needs long-term storage and can only be used once. Whether it has a long storage life determines its application prospect. Therefore, it has always been a hot topic in the industry that how to monitor and predict the health status of Solid Rocket Motor. Digital twin provides a new idea and way for the design and operation of complex equipment system with its characteristics of virtual reflecting reality, virtual controlling reality and virtual reality interaction. Aiming at the problem of health monitoring and failure prediction of solid rocket motor, this paper analyzes the factors that may cause the failure of solid rocket motor from the aspects of shell, grain, adhesive interface and nozzle. The architecture of health monitoring and failure prediction system based on digital twin is designed for solid rocket motor, and a four-dimensional digital twin model of solid rocket motor is constructed, which includes engine entity, digital twin model, engine database and control system. The data from entity to model and information from model to entity are transmitted between the engine entity and the twin model through the digital thread. The method of health monitoring and failure prediction of solid rocket motor based on digital twin is described in detail, in order to provide a new method for the storage and maintenance of solid rocket motor under the new technology conditions.