Method and system for predicting dynamic temperature of transformer

The invention discloses a transformer dynamic temperature prediction method and system. The method comprises the following steps: selecting an updated machine learning model corresponding to an optimal performance index as a black box model; adopting the black box model and a preset steady-state prediction model to generate a transformer thermal characteristic simulation model; and inputting the real-time operation data corresponding to the transformer information into a black box model or a steady-state prediction model corresponding to the transformer thermal characteristic simulation model according to a preset scene, and outputting dynamic temperature distribution. The technical problem that it is difficult to monitor the temperature of the iron core and the coil due to the fact that installation of thermometers on the iron core and the coil is hindered in the prior art is solved. Based on a lightweight technology and a multi-physics field simulation real-time calculation technology, the complexity of the