Prediction of End Point Sulfur of Hot Metal KR Desulfurization Based on IRLM Two-Step Prediction

In order to improve the accuracy of endpoint sulfur prediction and improve production efficiency, this paper proposes a two-step prediction model of hot metal endpoint sulfur based on Long Short-Term Memory(IRLM). Firstly, according to the correlation of model input data, Pearson correlation coefficient is used for feature selection to optimize the selection of model input parameters. Based on the improved Convolutional Neural Network - Long Short-Term Memory (CNN-LSTM) network, the model predicts the endpoint sulfur content and desulfurization rate by introducing the IRLM two-step prediction method. The experimental results show that compared with the traditional CNN-LSTM model, the prediction performance of the proposed model is significantly improved. The R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> value is increased by 12.64 %, the mean absolute error (MAE) is reduced by 50 %, and the root mean square error ( RMSE ) is reduced by 71 %. These results verify the accuracy and reliability of the model in complex industrial environments.