Static prediction of BOF steel-making end-point using an improved TSVR model based on BES optimization and fuzzy weighting
Abstract
To support the reduction of energy consumption in steel-making, accurate end-point prediction in steel-making is essential. This prediction enables precise process control, thereby reducing energy and material consumption. Conventional models often lack accuracy or robustness. Their applicability is limited under complex industrial conditions. This study proposes a new static prediction method. The method uses a bald eagle search-optimized fuzzy weighted twin support vector regression model. This model extends the baseline twin support vector regression. A fuzzy weighting mechanism is incorporated. The bald eagle search algorithm optimizes its hyperparameters. The key input variables are first selected with the gray relational analysis. The principal component analysis then extracts features and reduces noise. Each sample receives a fuzzy membership weight. This weight uses residual information from a base model. The processed features and weights integrate into the final model. This reduces the influence of outliers. The tests used 100 industrial datasets. The method outperforms traditional models in accuracy and stability. For carbon, the allowable error is ±0.005.wt.%. The hit rate reaches 88 %. For temperature, the allowable error is ±10 °C. The hit rate reaches 94%. Compared with the multilayer perceptron model, the hit rate for carbon content is improved by 9 percentage points. Compared with the linear regression model, the hit rate for temperature is improved by 6 percentage points. The proposed model is robust and self-optimizing. It provides a reliable reference for sensing key process parameters. High-accuracy prediction reduces reblowing operations. It minimizes molten steel over-oxidation. Oxygen and alloy consumption are also lowered. This work directly supports energy and material conservation. It aids the green transformation of basic oxygen furnace steel-making. The study has clear significance for sustainable metallurgical practice.