Intelligent Prediction of Oxygen Consumption in Steelmaking Based on Random Forest Method

The prediction of oxygen consumption during the steel smelting process is crucial for enhancing production efficiency and optimizing resource utilization. This study aims to propose a novel combination of data processing and model construction for predicting oxygen consumption in steelmaking. Firstly, we collected data related to the steel smelting process and employed effective data preprocessing techniques. Utilizing Permutation Importance for feature selection, we established a dataset comprising features such as temperature, molten iron and scrap metal mass, various chemical components, and more. Subsequently, we used a random forest model for training and ultimately employed the model to predict oxygen consumption. The results indicate that the random forest model exhibits excellent performance in predicting oxygen consumption during steel smelting. Compared to other deep learning models or complex alternatives, this model features more effective feature selection. With fewer input dimensions, the model demonstrates faster computational speed while achieving prediction accuracy nearly equivalent to complex models. Additionally, this study provides insights into the crucial factors influencing oxygen consumption, offering valuable references for optimizing the steel smelting process. We hope to contribute to the steel industry by presenting an effective method for predicting oxygen consumption, thereby enhancing production efficiency and resource utilization.