Method and apparatus for rolling a metal strip

A method for rolling a metal strip in a roughing line and a finishing line, wherein a rolling process in the roughing line is adjusted as a function of a predicted value for a change in width of the metal strip in the finishing line such that the metal strip has a predetermined finished strip width when leaving the finishing line, the method comprising:

• calculating first influencing parameters before a last edging pass in the roughing line to preset the finishing line, the change in width being dependent on the first influencing parameters;
• sending the first influencing parameters and variable network parameters to a neural network;
• generating, using the neural network, a calculated value for the change in width as a function of the predicted value;
• measuring and calculating second influencing parameters after the metal strip passes through the finishing line as a function of other measured process parameters;
• providing the second influencing parameters to the neural network after the strip metal passes through the finishing line;
• and
• reducing a deviation between a resulting network response and a measured change in width of the metal strip by adapting the variable network parameters of the neural network, wherein, when a type of metal strip is changed, the resulting network response is used as the predicted value for the change in width and to adjust the roughing line, and wherein, when processing successive metal strips of a same type, the measured change in width of the preceding successive metal strips is used as the predicted value for the change in width and for adjusting the roughing line.

A method for rolling a metal strip in a roughing line and a finishing line, wherein a rolling process in the roughing line is adjusted as a function of a predicted value for a change in width of the metal strip in the finishing line such that the metal strip has a predetermined finished strip width when leaving the finishing line, the method comprising:

• calculating first influencing parameters before a last edging pass in the roughing line to preset the finishing line, the change in width being dependent on the first influencing parameters;
• sending the first influencing parameters and variable network parameters to a neural network;
• generating, using the neural network, a calculated value for the change in width as a function of the predicted value;
• measuring and calculating second influencing parameters after the metal strip passes through the finishing line as a function of other measured process parameters;
• providing the second influencing parameters to the neural network after the strip metal passes through the finishing line;
• and
• reducing a deviation between a resulting network response and a measured change in width of the metal strip by adapting the variable network parameters of the neural network, wherein, when at least one strip is positioned between a first metal strip (n+i, where i.gtoreq.1) upstream from at least the last edging pass in the roughing line and a second metal strip (n) leaving the finishing line, the predicted value for the change in width of the first metal strip in the finishing line is calculated from a calculated value (y.sub.NN (x.sub.prev (n+i))) generated by the neural network, the resulting network response (y.sub.NN (x.sub.after (n))) being generated for the first metal strip leaving the finishing line based on the second influencing parameters that are one of calculated and measured, and wherein the measured change in width y.sub.actual (n) is calculated according to the equation:
• y.sub.pre (n+1)=K.sub.1 .multidot.y.sub.actual (n)+K.sub.2 .multidot.y.sub.NN (n+1))+(1-K.sub.1 -K.sub.2).multidot.y.sub.NN (x.sub.after (n)).
• where 0.ltoreq.k.sub.1 and k.sub.2.

An apparatus for rolling a metal strip in a roughing line and a finishing line, wherein a rolling process in the roughing line is adjusted as a function of a predicted value for a change in width of the metal strip in the finishing line such that the metal strip has a predetermined finished strip width when leaving the finishing line, the apparatus comprising:

• a control unit calculating first influencing parameters before a last edging pass in the roughing line to preset the finishing line, the change in width being dependent on the first influencing parameters;
• a neural network having variable network parameters and receiving the first influencing parameters, the neural network generating a calculated value for the change in width as a function of the predicted value;
• and
• a measurement logger for measuring and calculating second influencing parameters after the metal strip passes through the finishing line as a function of other measured process parameters,
• wherein, after the strip metal passes through the finishing line, the second influencing parameters are provided to the neural network, and
• wherein the neural network reduces a deviation between a resulting network response and a measured change in width of the metal strip by adapting the variable network parameters of the neural network.

An apparatus for rolling a metal strip in a roughing line and a finishing line, wherein a rolling process in the roughing line is adjusted as a function of a predicted value for a change in width of the metal strip in the finishing line such that the metal strip has a predetermined finished strip width when leaving the finishing line, the apparatus comprising:

• a control unit calculating first influencing parameters before a last edging pass in the roughing line to preset the finishing line, the change in width being dependent on the first influencing parameters;
• a neural network having variable network parameters and receiving the first influencing parameters, the neural network generating a calculated value for the change in width as a function of the predicted value;
• and
• a measurement logger for measuring and calculating second influencing parameters after the metal strip passes through the finishing line as a function of other measured process parameters,
• wherein, after the strip metal passes through the finishing line, the second influencing parameters are provided to the neural network,
• wherein the neural network reduces a deviation between a resulting network response and a measured change in width of the metal strip by adapting the variable network parameters of the neural network, and
• wherein, when a type of metal strip is changed, the resulting network response is used as the predicted value for the change in width and to adjust the roughing line, and wherein, when processing successive metal strips of a same type, the measured change in width of the preceding successive metal strips is used as the predicted value for the change in width and for adjusting the roughing line.

An apparatus for rolling a metal strip in a roughing line and a finishing line, wherein a rolling process in the roughing line is adjusted as a function of a predicted value for a change in width of the metal strip in the finishing line such that the metal strip has a predetermined finished strip width when leaving the finishing line, the apparatus comprising:

• a control unit calculating first influencing parameters before a last edging pass in the roughing line to preset the finishing line, the change in width being dependent on the first influencing parameters;
• a neural network having variable network parameters and receiving the influencing parameters, the neural network generating a calculated value for the change in width as a function of the predicted value;
• and
• a measurement logger for measuring and calculating second influencing parameters after the metal strip passes through the finishing line as a function of other measured process parameters,
• wherein, after the strip metal passes through the finishing line, the second influencing parameters are provided to the neural network,
• wherein the neural network reduces a deviation between a resulting network response and a measured change in width of the metal strip by adapting the variable network parameters of the neural network, and
• wherein, when at least one strip is positioned between a first metal strip (n+i, where i.gtoreq.1) upstream from at least the last edging pass in the roughing line and a second metal strip (n) leaving the finishing line, the predicted value for the change in width of the first metal strip in the finishing line is calculated from a calculated value (y.sub.NN (x.sub.prev (n+i))) generated by the neural network, the resulting network response (y.sub.NN (x.sub.after (n))) being generated for the first metal strip leaving the finishing line based on the second influencing parameters that are one of calculated and measured, and wherein the measured change in width y.sub.actual (n) is calculated according to the equation:
• y.sub.pre (n+1)=k.sub.1 .multidot.y.sub.actual (n)+k.sub.2 .multidot.y.sub.NN (x.sub.prev (n+1))+(1-k.sub.1 -k.sub.2).multidot.y.sub.NN (x.sub.after (n)),
• where 0.ltoreq.k.sub.1 and k.sub.2 .ltoreq..