With the development of information technology, digital twin technology has become a new form of technology in the field of condition monitoring, and also provides a new solution for junction temperature estimation. This paper takes digital twin as the starting point, aiming at the problems existing in the current junction temperature monitoring method, explores the feasible junction temperature monitoring method of power module. From the point of view of the internal structure mechanism of the module, the mode and way of heat transfer in the module are explored, and the factors affecting the junction temperature of the module are analyzed. The finite element modeling was carried out based on the real structure of the module, and the input variables of the twin model were determined through the data collected by the controller. The training set of DOE experiment was designed, and the finite element model was trained and reduced by Ansys twinbuilder to obtain the twin model of the power module, and the experimental platform was built to verify the feasibility of the model.