Study On Principle Of Flying Gauge Change And Rolling Force Preset For Tandem Cold Mill

Flying Gauge Change (FGC) is not only the exclusive process of fully continuous tandem cold mills (TCM) but also the key technique to realize strip continuous rolling. Therefore, research on FGC in TCM is vital important to increase of its yield, improvement of its product quality, etc. Precision of rolling force preset is an important factor affecting the precision of thickness and flatness of coiling, so, improving the precision of rolling force preset is an important problem of FGC.To meet the increasing requirements of strip thickness and flatness accuracy, it is becoming more and more important to improve the precision of presetting rolling parameters. In this thesis it is focused on studying the rolling force preset because the rolling force is the most important factor influencing strip thickness and flatness. The chief research work are constituted by three parts as follows:Part one mainly analyzes and studies current technique of FGC in TCM. Above all, the process of FGC was briefly explained so that the questions and their demands could be clearly understood.The second part studies on the adaptive learning models of deformation resistance and friction which are the most important factors influencing rolling force. The backward calculation models of deformation resistance model and friction model based on actual rolling force and actual forward slip in order to improve the rolling force model calculation accuracy. Further, the adaptive learning models of deformation resistance and friction are built up based on backward calculation models.For further improvement of rolling force accuracy, Artificial Intelligence (mostly includes neural network , GA (Genetic Algorithm) and combine with rolling force mathematic model) is introduced to this study. After analysis of each characters of BP (back-propagation) algorithm and GA algorithm, against the BP neural network's deficiency of easily falling into local minimum, we propose to use an algorithm combined GA with BP to overcome BP neural network's inherent deficiency. After analysis of the feasibility of algorithm combined GA with BP, we explore how to...