Comprehensive status evaluation and prediction of blast furnace based on cascade system and combined model

• A dual cascade evaluation system integrating subjective and objective weighting methods enables accurate quantitative scoring of blast furnace comprehensive status. • The combined BiLSTM and CatBoost model significantly improves prediction accuracy, reducing mean absolute error by 0.275 and increasing hit rate by 5.65 percentage points on average. • With an error range of ±2.5, the combined model achieves a 91.66% hit rate for next-hour comprehensive status prediction, demonstrating field-ready reliability. • SHAP analysis reveals a linear relationship: a 10°C increase in furnace bottom center temperature raises the comprehensive status score by 0.44, offering actionable operational insights.