Effect of Basicity on the Microstructure of Sinter and Its Application Based on Deep Learning

Area of Science:

• Metallurgical Engineering
• Materials Science

Background:

• Sinter quality is crucial for ironmaking efficiency.
• Basicity is a key parameter influencing sinter mineralogy and properties.

Purpose of the Study:

• To investigate the effect of basicity (0.6–2.4) on sinter mineral composition and microstructure.
• To correlate mineralogical changes with sinter quality indicators like the drum index, vertical sintering speed, and yield.

Main Methods:

• Polarizing microscopy was used to analyze mineral composition and microstructure.
• Comprehensive analysis of drum index, vertical sintering speed, and yield was performed.
• A convolutional neural network was applied for predicting basicity from mineral facies images.

Main Results:

• Basicity increase from 0.6 to 1.0 shifted mineral structure from porphyritic-granular to porphyritic, decreasing quality indicators.
• At basicity ~1.6, an interweaving corrosion structure emerged, but inhomogeneity and low drum index were observed.
• Optimal sinter quality, indicated by the highest drum index, was achieved at a basicity of 2.0, with a uniform interweaving corrosion structure and maximum calcium ferrite content.
• Basicity above 2.0 led to a slight decline in quality indicators.

Conclusions:

• Sinter production should control basicity around 2.0, avoiding levels near 1.0 and 1.6.
• Mineral facies analysis combined with machine learning (convolutional neural networks) offers a novel approach for predicting basicity and optimizing sinter production.