Power grid frequency situation prediction method based on transfer learning

The invention discloses a power grid frequency situation prediction method based on transfer learning, which can adapt to frequency situation prediction requirements when a power grid operation mode and a topological structure are changed, and improves accuracy and reliability of a model. The method comprises the following steps: firstly, constructing a post-fault frequency situation prediction model based on a convolutional neural network (CNN), then obtaining enough samples for training by utilizing a transfer learning method, and correcting parameters of the CNN frequency situation prediction model, thereby obtaining a more accurate prediction model, and improving the precision of a system frequency prediction result. The precision of the frequency situation prediction model can be remarkably improved, the frequency change situation of the disturbed system can be predicted more accurately, corresponding control measures such as generator tripping, load shedding and direct-current emergency power control ar