Steelmaking Process: Neural Models Improve End-Point Predictions

摘要

The basic oxygen steelmaking (BOS) is a transient process, highly complex and is also subject to oscillations in raw material composition. A robust model is essential to adjust end-blow oxygen and coolant requirements to match with the targets of end-point temperature and carbon percentage in liquid steel. This paper describes the development of neural models and the industrial application to the BOS plant of the Cia. Siderurgica Nacional (CSN-Volta Redonda/Brazil). The inverse neural model is responsible for end-blow process adjustments. At the end of 40 industrial runs, it was obtained 82.5% of simultaneous agreement with the targets, against 66% obtained from the commercial model currently used at CSN's plant. End-point temperature goal was achieved in 97.5% of the cases through the neural model corrections. The performance improvement shows that the neural model is a potential tool to automatically control the BOS process.