Blast furnace tuyere working state identification method

The invention particularly relates to a blast-furnace tuyere working state identification method, which sequentially comprises the following steps of: manufacturing a blast-furnace tuyere working state identification model data set, inputting the manufactured data set into a blast-furnace tuyere working state identification model to carry out convolutional neural network forward propagation, generating a gradient weighting type activation mapping graph, and carrying out gradient weighting type activation on the gradient weighting type activation mapping graph. The method comprises the following steps of: performing up-sampling on a gradient weighting type activation mapping graph to obtain a gradient weighting type activation mapping graph with the same size as an image of an original input data set, and guiding clipping data enhancement and discarded data enhancement according to the gradient weighting type activation mapping graph; and jointly inputting the image of the original input data set and the gradi