Narrow Results

Customers for cold rolled strip products expect the good flatness and surface finish, consistent metallurgical properties and accurate strip thickness. These requirements demand accurate prediction model for rolling parameters. This paper presents a set-up optimization system developed to predict the rolling force during cold strip rolling. As the rolling force has the very nonlinear and time-varying characteristics, conventional methods with simple mathematical models and a coarse learning scheme are not sufficient to achieve a good prediction for rolling force. In this work, all the factors that influence the rolling force are analyzed. A hybrid mathematical roll force model and an adaptive neural network have been improved by adjusting the adaptive learning algorithm. A good agreement between the calculated results and measured values verifies that the approach is applicable in the prediction of rolling force during cold rolling of thin strips, and the developed model is efficient and stable.

In order to analyze the effects of cold rolling parameters such as the crossing angle and axial shifting value of work rolls on the strip profile, extensive tests were carried out on a 4-high rolling mill equipped with a work roll crossing and shifting system. The results show that the strip profile is nearly flat under asymmetrical rolling. The rolling force was also analyzed in detail by changing the crossing angle and axial shifting value of work rolls.

Customers for cold rolled strip products expect the good flatness and surface finish, consistent metallurgical properties and accurate strip thickness. These requirements demand accurate prediction model for rolling parameters. This paper presents a set-up optimization system developed to predict the rolling force during cold strip rolling. As the rolling force has the very nonlinear and time-varying characteristics, conventional methods with simple mathematical models and a coarse learning scheme are not sufficient to achieve a good prediction for rolling force. In this work, all the factors that influence the rolling force are analyzed. A hybrid mathematical roll force model and an adaptive neural network have been improved by adjusting the adaptive learning algorithm. A good agreement between the calculated results and measured values verifies that the approach is applicable in the prediction of rolling force during cold rolling of thin strips, and the developed model is efficient and stable.