Estimation of plate temperatures in hot rolling based on an extended Kalman filter

An integrated model and an extended Kalman filter (EKF) of the product temperatures in hot rolling are presented. The dynamical model takes into account the effect of all significant production steps on the temperature evolution of the plate, i.e., roll passes, descaling passes, and air cooling periods. Consequently, the model structure is switching. Based on this model, an EKF is developed that uses surface temperature measurements from pyrometers installed along the production line. However, these pyrometers measure only at discrete points in time, when the plate passes the pyrometer. Thus, the EKF has to cope with the switching model structure and with scarce measurements. To test the performance of the developed EKF, experiments with supplementary pyrometer measurements were carried out so that the observed temperature can be compared to measurements which were not used as inputs of the filter.