A Hybrid Ensemble Model Based on ELM and Improved AdaBoost.RT Algorithm for Predicting the Iron Ore Sintering...

Area of Science:

• Metallurgical Engineering
• Computational Intelligence

Background:

• The steel industry faces increasing pressure to improve energy efficiency.
• Iron ore sintering is a significant energy consumer in iron and steel production.
• Accurate prediction of sintering characteristics is crucial for process optimization.

Purpose of the Study:

• To develop a hybrid ensemble model for predicting key iron ore sintering characters.
• To enhance energy efficiency and sinter quality in the iron ore sintering process.
• To identify and rank factors influencing sintering performance.

Main Methods:

• Utilized a hybrid ensemble model combining Extreme Learning Machine (ELM) with an improved AdaBoost.RT algorithm.
• Employed the RReliefF method for ranking factors affecting solid fuel consumption, gas fuel consumption, burn-through point (BTP), and tumbler index (TI).
• Developed an improved AdaBoost.RT algorithm with dynamic threshold adjustment for enhanced prediction accuracy.

Main Results:

• The hybrid ensemble ELM model effectively predicted iron ore sintering characters using production data.
• The RReliefF method successfully identified key factors influencing sintering performance.
• The proposed model demonstrated feasibility and effectiveness for practical sintering processes.

Conclusions:

• The developed hybrid ensemble model offers a robust solution for predicting iron ore sintering characters.
• Analysis of superior factors provides insights for improving energy efficiency and sinter quality.
• The model's success in a real-world industrial setting (Baosteel) validates its practical applicability.