| 摘要 |
PURPOSE: A method for improving prediction accuracy of roll force in hot rolling by introducing a method for gradually adapting learning in the same section as well as roll force learning coefficient of neighboring records is provided. CONSTITUTION: In a method for predicting roll force in hot rolling for the same rolling material records having constituents, thickness and width of steel as factors during hot rolling of steel or a plurality of rolling material records divided by the factor, the method comprises a step of obtaining adjacent records; a step of obtaining Grf correlation coefficient between neighboring records for the obtained adjacent records; a step of obtaining Kαβγ; a step of obtaining Grf,n,i+α j+β k+τ by using the obtained Grf correlation coefficient and Kαβτ; and a step of predicting roll force by the following mathematical expression 9 using the obtained Grf,n,i+α j+β k+τ:£Mathematical Expression 9|Fn+1=GRf,n,i+α j+β k+τx Fmodel,n+1.
|
