Abstract
Although scrap-based electric arc furnace (EAF) plays a pivotal role in achieving net-zero waste and emission targets in the steel industry, this technology is not without scientific and technical challenges. For example, the diversity of scrap feed in terms of composition requires an adjustment of operational variables considering the final steel quality such as the concentration of harmful impurities. In addition, relying solely on empirical methods can be very time-consuming and costly. In recent years, neural networks and more specifically deep learning have attracted the attention of researchers in steelmaking. A combination of a metallurgical understanding of a process and artificial intelligence capabilities makes process optimization time- and cost-effective. Therefore, in this research, a model based on principal component analysis (PCA) and recurrent neural network (RNN) is proposed to predict the phosphorus content of steel produced in a scrap-based EAF considering practiced process parameters (e.g., steel composition, temperature, slag formers, injected oxygen, etc.). Subsequently, the combined PCA-RNN model is tested against the historical data of the steel plant. Further details of this study are discussed in the present work. Although scrap-based electric arc furnace (EAF) plays a pivotal role in achieving net-zero waste and emission targets in the steel industry, this technology is not without scientific and technical challenges. For example, the diversity of scrap feed in terms of composition requires an adjustment of operational variables considering the final steel quality such as the concentration of harmful impurities. In addition, relying solely on empirical methods can be very time-consuming and costly. In recent years, neural networks and more specifically deep learning have attracted the attention of researchers in steelmaking. A combination of a metallurgical understanding of a process and artificial intelligence capabilities makes process optimization time- and cost-effective. Therefore, in this research, a model based on principal component analysis (PCA) and recurrent neural network (RNN) is proposed to predict the phosphorus content of steel produced in a scrap-based EAF considering practiced process parameters (e.g., steel composition, temperature, slag formers, injected oxygen, etc.). Subsequently, the combined PCA-RNN model is tested against the historical data of the steel plant. Further details of this study are discussed in the present work.