Adaptive Weighted Broad Echo State Learning System-Based Dynamic Modeling of Carbon Consumption in Sintering Process.

Area of Science:

• Metallurgical Engineering
• Industrial Process Control
• Machine Learning Applications

Background:

• Dynamic modeling of carbon consumption is crucial for optimizing the iron ore sintering process.
• Energy saving, emission reduction, and green manufacturing goals necessitate accurate carbon consumption prediction.

Purpose of the Study:

• To propose a novel adaptive weighted broad echo state learning system (AWBESLS) for dynamic carbon consumption prediction.
• To enhance the accuracy and robustness of predictions by addressing anomalous production data.

Main Methods:

• The AWBESLS integrates adaptive weights with an echo state network (ESN).
• Adaptive weights are assigned to production data to mitigate the impact of anomalies.
• An ESN is employed to capture the dynamic states inherent in the sintering process.

Main Results:

• Experiments using actual production data demonstrate the effectiveness of the AWBESLS.
• The AWBESLS significantly outperforms existing state-of-the-art methods in prediction accuracy.
• The proposed system achieved the lowest prediction error among compared methods.

Conclusions:

• The AWBESLS is an effective and applicable technique for the dynamic modeling of the sintering process.
• The methodology shows promise for application in modeling other complex manufacturing processes.
• This approach contributes to more sustainable and efficient industrial operations.