Problems of modelling of meld stress in the on-line control of hot rollingprocesses

Model of the yield stress, applicable to the on-line control of hot steel p late rolling, is described in the paper. Developed model has a physical mea ning. Three methods, based on an analysis of large number of experimental d ata, are applied for evaluation of coefficients in the model. These methods are approximation, optimisation and artificial neural network. Yield stres s was not measured directly. It was determined from the roll force measurem ents, using inverse calculations of Sims equation. The results for three gr oups of steels (two carbon-manganese steels and one niobium steel), are pre sented in the publication. The adaptation technique is described in the pap er, as well. Application of the adaptation technique allows for immediate r eaction of the system on the changes of yield stress. Experimental validati on of the model confirmed its good accuracy and usefulness for the on-line control of the hot plate rolling process.