PURPOSE: To offer a device for automatically recognizing and controlling shape of band steel with which the shape of band steel during rolling can be improved.
CONSTITUTION: The controlled variable of actuator is corrected with a controller 10 for control based on a recognized shape that is recognized with a recognition neural net 5 and leveling control of hot finishing mill is feedback controlled at real time, on the other hand, simultaneously the set value for setting bender of the rolled stock which is next rolled is set with a computer 9 for control based on the estimated shape of band steel that is estimated with an estimating neural net 8 and outputted to the controller 10 for control. In the case of the deterioration of the shape of band steel, feedback control for improving it can be done at real time, simultaneously the shape of band steel which is next formed is estimated and this can be reflected to the initialization for controlling rolling work. Highly accurate control is made possible by adopting the neural net using error inverse propagation method and the whole of band steel can be always made into a good shape.
ITO TAKA
AIHARA KAZUYUKI
