Blast furnace a testimony to Baosteel's smart AI use

Previously, furnace temperature was managed mainly through the experience of the company's senior operators.

"With large AI models the blast furnace, which was once a 'black box', has become far more transparent. By analyzing complex internal interactions, the system can now generate highly accurate forecasts, with predictions made two hours ahead and exceeding 90 percent accuracy," Wang said, adding that work efficiency has also improved by 20 percent.

"Taking the blast furnaces at the Baoshan base for example, the target is to maintain an ideal range of 1,510, plus or minus 10 degrees Celsius, with longer stability providing stronger operational assurance," he added.

The company officially launched the world's first AI-powered smart blast furnace model on April 3.

This marked that Baosteel has leveraged AI large models to improve furnace temperature predictions, providing a practical model for the steel industry's digital transformation by converting experience into data and uncertainty into precision.

Blast furnace ironmaking accounts for around 70 percent of total steel production costs. In this process, temperature fluctuations are not only cost-related, but also closely linked to carbon emissions.

According to Wang, the AI large model has played a significant role in ensuring stable blast furnace operations. Prediction accuracy has improved from just over 70 percent at the beginning of 2024 to over 90 percent, and continues to be optimized. Other external production bases have invited the team to help deploy the system.

Before embracing AI large models, Baosteel had already built a strong basis for using cutting-edge technologies.

Since the 1990s, the steelmaker has focused on information system development and experience accumulation, and further advanced digitalization to form a sizable database.