PREDICTIVE CONTROL OF METAL COOLING ON A CONTINUOUS BILLET CASTING MACHINE WITH SURFACE TEMPERATURE MONITORING

The paper proposes a control system for secondary metal cooling on continuous billet casting machines, using surface temperature measurements between cooling zones. The aim is to maintain a prescribed surface temperature at the outlet of each secondary cooling zone by adjusting the cooling water flow. The temperature dynamics along the zones is assumed to be known from a model and specified a priori. A model of the control system is developed to study con-trol features. The work shows that, due to time delay, a conventional controller responds unsatisfactorily to disturb-ances in metal temperature at the section inlet. To compensate for disturbances in real time, the control system is proposed to account for the metal temperature at the inlet to the secondary cooling zone. The work shows that using a repetitive controller can completely eliminate the effect of periodic disturbances inherent to the plant, provided accu-rate information on the current parameters of a simplified model is available. The paper proposes a new-type identifi-cation system to determine the parameters of the simplified model. During identification, the parameter characterizing unknown disturbances should minimize the scatter of stationary model parameters estimated over a given time interval. For the problem at hand, a stationary parameter of the simplified model (for the current operating mode) is understood as the ratio of the surface temperature increment to the change in water flow, while the parameter characterizing unknown disturbances is the water flow required to keep the surface temperature constant. Given the effectiveness of the proposed system in disturbance compensation, it can be used together with interpretable metal cooling models that define the desired surface temperature trajectory and controller settings. The proposed identification system allows the interpretable model to be adapted under uncertainty of the surface heat flux and in the absence of information on the temperature distribution inside the billet. The repetitive controller, the interpretable model, and the identification system together form a self-tuning metal temperature control system.