Mathematical modeling and observer design for the plate temperature in hot rolling

Since steel is still in demand as a strong and versatile material, ongoing research efforts try to improve the quality of steel products and the efficiency of the production processes. This thesis deals with the hot rolling of heavy plates and focuses on the evolution of the plate temperature because its good understanding is essential to plan the rolling process. First, a first-principles mathematical model is developed that yields a plate temperature profile through the thickness and a radial temperature profile of the work rolls during the production process. The thermal model is parametrized and validated using measurement data. Additionally, a computationally inexpensive and sufficiently accurate reduced model is deduced. This model considers the work rolls only during the short contact times with the plate. Second, an Extended Kalman Filter (EKF) is designed based on the full model as an observer of the plate temperature. The EKF corrects the estimated temperatures using rare and infrequent pyrometer measurements of the surface temperatures, which are the only available temperature measurements under normal production conditions. As the performance of the EKF is determined by its initialization and parametrization, these aspects are discussed in detail. Despite the scarcity of available measurements, the estimated temperatures agree well with supplementary measurements recorded during a verification experiment. Furthermore, a second EKF based on the reduced model is deduced. Both observers are comparably accurate, but the second one is computationally less expensive and can be used in real-time applications. Finally, this thesis also provides an analysis of the existing measurement facilities in view of possible measurement errors.