Health prediction and digital twin are essential to autonomous control and health management of satellites in orbit. They support small satellites’ autonomous control and ensure their safe and reliable in-orbit operation. For small satellites’ characteristics of low digitalization， low data flows and complex and variable operation environments， this paper proposes a framework of intelligent diagnosis and health prediction for the whole life cycle of small satellites based on digital twin. Firstly， an overview of health prediction， digital twin， digital mainline and their applications in the satellite field is provided. Secondly， the digital twin models for small satellite health prediction and a digital mainline for the whole life cycle of small satellites are constructed by making full use of various data. Finally， combining with the mechanism and performance data， the weak process identification， intelligent diagnosis， health assessment and remaining life prediction methods based on information fusion are proposed. The proposed framework can effectively realize physical and virtual data fusion， virtual-real interconnection， and data processing to achieve accurate health prediction of small satellites.