Data mining and model-predictive approach for blast furnace thermal control

This research proposes a method of blast furnace control based on criteria of increased productivity and lowers coke consumption. The method employs model-predictive control technology. Herewith constructing the model of the blast furnace process involves real-time operating regime data. Model-building assumes two approaches for clustering of operating parameters values using criteria of blast furnace efficiency. The first one uses elliptic surfaces. The second employs self-organizing Kohonen networks. Moreover when having the lack of informative measurements data the solution of the first task is used to normalize the solution of the second task. The research sets and solves the problem of real-time optimization of the blast furnace regime parameters.