GIS partial discharge type identification method and system based on transfer learning

The invention discloses a GIS partial discharge type identification method and system based on transfer learning. The method comprises the following steps: obtaining a PRPD atlas, and inputting the PRPD atlas into a preset GIS partial discharge type identification model to obtain a discharge type identification result; the construction process of the GIS partial discharge type identification model comprises the following steps: training a teacher network model on an ImageNet data set to obtain training feature parameters, migrating the training feature parameters to a student network model through transfer learning, and adding an attention mechanism to an input layer and an output layer of the student network model to obtain an initial identification model; preprocessing the PRPD sample map to obtain a training sample image, and repeating the iterative training process until a training loss value is obtained to obtain a GIS partial discharge type identification model; according to the method, the generalizati