Application of ANN for the dimensional accuracy of workpiece in hot rod rolling process

Abstract

A theoretical analysis is performed, which adjust the roll gap (pass height) to keep the dimension accuracy of exit cross-sectional shape of workpiece in rod (or bar) rolling process, with roll wear considered. The Archard’s wear model was modified as an incremental form and then wear depth of roll was calculated at each deformation step on contact area using finite element method coupled with modified wear model. The thermal softening of work roll during rolling was expressed in terms of the main tempering curve. Artificial neural network (ANN) has then been applied to the proposed model to keep the cross-sectional area constant during rolling which inevitably involves wear. Results showed that the proposed roll wear model and application method might give more systematically and economically feasible means to improve the dimension accuracy of exit cross-sectional shape of workpiece in rod (or bar) rolling process, by adjusting roll gap.