The invention discloses an aluminum electrolysis superheat degree recognition method based on CNN feature fusion and a semi-supervised Laplace extreme learning machine, and the method comprises the steps: 1, collecting aluminum electrolysis real-time production data, and carrying out the normalization and standardization of the collected data; 2, utilizing a convolutional neural network (CNN) to extract depth features of a fire eye image in the aluminum electrolysis industrial process; 3, fusing the depth features of the fire eye image extracted in the step 2 with other features of the fire eye image; and 4, constructing a semi-supervised extreme learning machine (LapsELM) as a classifier by using LapsELM regularization, and judging the current state of the superheat degree of the electrolytic cell according to the fire eye image.