Research on Prediction Model for End-point and Product Composition of Copper Converter Smelting Based on CNN-GAT Algorithm Collaboration

The end-point judgment of copper converter blowing directly affects the quality of blister copper, the stability of the furnace condition, and the blowing efficiency. Therefore, enhancing the digitalization and intelligence level of this process holds significant practical importance. This study employs a deep learning algorithm that integrates CNN (Convolutional Neural Network) and GAT (Graph Attention Network). It utilizes CNNs to extract image features from cooling samples of high-temperature melts. Subsequently, by fusing these image features with various production condition data and constraints through GAT, a model is constructed for endpoint judgment and prediction of product composition content. This model can predict the main element content of the furnace products and estimate the required blowing time. A dataset comprising 5,172 production parameters and images of high-temperature cooling samples from the furnace was established. The model was trained and validated using this dataset, and the results indicate that the model achieved endpoint judgment accuracies of 96.73% and 97.85% for the slag-making and copper-making periods, respectively, on the test set. The average prediction error for the composition content across four cycles of copper converter blowing was as low as 0.705%, and the average error in estimating the required blowing time was only 1.94 minutes. This study provides new methods and insights for the development of intelligent endpoint judgment technologies for copper converter blowing.