Electric power area meteorological disaster prediction method and system based on convolution algorithm

The invention discloses a power region meteorological disaster prediction method and system based on a convolution algorithm, and relates to the technical field of meteorological prediction, and the method comprises the steps: determining meteorological data in a meteorological station intersection region and an unknown region; dividing the power region into a plurality of sub-region graphs according to the geographic features of the power region graph, calculating the complexity level of meteorological data in the sub-region graphs, classifying the sub-region graphs according to the complexity level, and endowing each type of sub-region graphs with a target grid size; determining the influence attribute of each meteorological disaster to establish a meteorological disaster grade evaluation standard; configuring model structure parameters through meteorological disaster grades, and training a convolutional LSTM model; and predicting meteorological disasters by using the electric power region meteorological di