Prediction Model for Viscosity of Titanium-Bearing Slag Based on the HIsmelt Process

With the development of “Industry 4.0”, artificial intelligence technology is gradually being applied in the steel industry. In actual production, there is a serious lag in viscosity measurement methods. First, this paper used big data technology to analyze and process the viscosity test data and then established prediction models based on the CatBoost model and the LSTM model, respectively, for the viscosity of the titanium-bearing slag. Finally, the information entropy method was used to integrate the results, and an integrated model was established to predict the viscosity of the titanium-bearing slag. The results showed that the integrated model had better performance than the other two single models, with a prediction accuracy of 96.04% and an improvement of 2.06–5.49% compared with the prediction accuracy of the single models. The results show that the integrated model has more practical significance in guiding production practices.