Page 287 - Distributed model predictive control for plant-wide systems

Hot-Rolled Strip Laminar Cooling Process with Distributed Predictive Control 261

temperature distribution of the strip. Then, the EKF sends current states to the correspond-
ing local MPCs. In each local MPC, neighborhood optimization is adopted to enhance the
global control performance. Furthermore, to overcome the computational obstacle of nonlin-
ear model, the prediction model of each MPC is linearized around the current operating point
at each step. Through this method, the online optimization of strip cooling curve is realized
with a few computational burdens; both simulation and experiment results proved the effi-
ciency of the proposed method. Following this method, the whole evolution procedure of strip
temperature is controlled online with a relatively high precision, which provides possibilities
of producing many new types of high-quality steel (e.g., the multiphase steel).
In HSLC, the aim is to obtain a uniform microstructure of the strip. Therefore, it is more
reasonable that the temperature errors among different strip-points caused by FT are eliminated
gradually along the rolling direction. However, in this work, these errors are eliminated mainly
by the first several header groups, which is a problem to be solved in further works.