Modeling and Optimization of Caliber Rolling Process

Abstract

Abstract In practice of caliber rolling, it is highly desirable to obtain optimal processing parameters to achieve the best quality in a short time. A sophisticated and faster simulation model of caliber rolling is proposed, and then a global optimal searching method—simulated annealing algorithm (SA)—is applied to find the optimal processing parameters of caliber rolling. The simulation model of caliber rolling in this paper was established by the back-propagation neural network (BPNN) model, which replaces the complex numerical analytical model, proposed in an earlier paper. This replacement improved the speed of simulation significantly.