Data collected by blast-furnace sensors

Area of Science:

• Metallurgical Engineering
• Process Control
• Data Science

Background:

• Blast-furnace smelting generates complex, high-dimensional data.
• Raw data often contains noise and outliers, requiring careful processing.
• Accurate technological parameters are crucial for efficient iron production.

Purpose of the Study:

• To provide validated, synchronized datasets of blast-furnace smelting parameters.
• To facilitate research in process monitoring and optimization.
• To enable the development of data-driven models for blast-furnace operations.

Main Methods:

• Data collection from dispatch systems.
• Data cleaning, validation, and outlier removal.
• Synchronization of time-series data recorded at hourly intervals.

Main Results:

• Two datasets were generated: raw, noisy data and validated, class-labeled data.
• Key parameters recorded include gas pressure, temperature, composition (CO2, H2), and blast conditions.
• Silicon content in hot metal was also included.

Conclusions:

• The validated datasets offer a reliable resource for blast-furnace research.
• Availability of clean, synchronized data supports advanced analytical techniques.
• This work contributes to improving the understanding and control of smelting processes.