Prediction and Control of the Mechanical Properties of Rolled Products Via Probabilistic Modeling Methods

Considering the mechanical properties of steel, this work deals with an automated calculation algorithm of the temperature regime of steel hot rolling on broadband mills. The algorithm is based on probabilistic Bayes networks that are designed for the compact expression of the joint distribution density of the rolling parameters, chemical composition, and mechanical properties of rolled steel. In automated mill control systems, such an algorithm can correct setups for the finishing rolling temperature and the coiling temperature for melting specific orders. Depending of the conditions, these parameters can be adjusted up to 200°C, thus providing ample opportunity for tuning the mechanical properties of the metal. Modern control systems for second level broadband mills provide individual setup temperature parameters for the rolling of each coil (plate). The use of this characteristic in an automated mode reduces the intragrade dispersion of the mechanical properties of rolled products and increases yield stability.