Abstract: In order to improve the width control precision in rough rolling process， an adaption method for optimizing parameters of the width spread model was proposed. The natural width spread coefficient and the dog-bone width spread coefficient in the width model were taken as the adaption parameter vectors， and the width adaption model objective function was founded based on the vectors. The Nelder-Mead method was used to search the optimal solution of the objective function， and the optimized parameters vectors were obtained which fulfilled the model precision of the width control process. The results showed that the adaptive method is successfully applied to the control system of one 850mm hot tandem mill. Compared with the traditional method， the width deviation between the actual measured ones and the predicted ones is reduced from 3.05mm to 1.28mm with the proposed method. The validity and precision of the proposed adaption model are verified， and the width quality is enhanced greatly.

Key words: hot tandem mill, rough rolling, objective function, width model, Nelder-Mead method