Optimization of Model Adaption Based on Multi-Samples in Hot Strip Mill

In order to improve the thickness precision, the model adaption method is of great concern which is a necessary component in the thickness control system. The sample concept is first taken into the model adaption in hot tandem mill, from which the learning coefficients are updated by a series of samples. Comparing to the single sample in traditional method, the new method can consider the variance in the length direction. And the source data is more comprehensive, the robustness is increased with flexible proportion coefficients. The thickness precision in stable production process is enhanced to 35μm. On-line result shows practical value and on-line application prospect.