Research On Temper Mill Online Control System Of1550mm Cold Rolling Strip

The temper mill is the key equipment of the continuous annealing line for cold-rolled strip steel.The rolling process of the temper mill is the last process of the continuous annealing line for cold-rolled strip steel.It is extremely important for eliminating the yield platform,improving the surface quality of the finished product and the product yield ratio.The process parameter control in the rolling process is a typical nonlinear and multivariable coupling control.Process engineers often pursue the optimal combination of process parameters after completing the basic production requirements.However,due to the nonlinear and multivariable coupling characteristics of process parameters,the traditional mathematical model and static table control model have poor control effect on process parameters.In order to solve the existing problems,this paper combines the mechanical model of the rolling process,the finite element model of the rolling process and the incremental learning algorithm to establish a new intelligent temper mill control system,and successfully applied it online,the main research contents are as follows:(1)Based on the temper mill equipment of a 1550 mm continuous annealing unit,a mathematical model of rolling force and tension calculation in the rolling process is established.Based on the big data analysis method,the variation of the optimal rolling force with the strip thickness and steel grade in the temper rolling process was studied.It is proved that it is difficult to set the accurate rolling force value in the static process table model which only considers steel grade and thickness.(2)The two-dimensional finite element model of the roll-strip coupling was established using the finite element software DEFORM,and the rolling process was simulated and analyzed.The effects of different strip thickness,tension,friction and rolling speed on the elongation and stress-strain state of the strip were analyzed.(3)Based on the characteristics of the rolling process,the IID3-KNN(Improved Iterative Dichotomiser 3 and K-Nearest Neighbor)algorithm is developed.The algorithm learns different process parameter changes based on the decision tree structure and achieves efficient increments by adding corresponding branch nodes and leaf nodes Learning,combined with the KNN algorithm to find the nearest neighbor node at each level of the decision tree to predict the process parameters of the new specification steel coil,and search from the root node to the leaf when self-learning optimizes and predicts the process parameters of the old specification steel coil.The high efficiency self-learning and process parameter prediction are realized,and the problems of low control precision,low self-learning efficiency and incremental learning difficulty of the traditional mathematical model and static table control model are improved.(4)An online control system for cold-rolled strip steel temper mill is studied and applied online.Based on the mechanical model of the rolling process and the massive historical data of the on-site leveling and rolling process,a strip temper mill control model that can be optimized by online incremental learning and self-learning is designed.The control system includes a data communication module,a data collection module,and a control module.Finally,based on the actual application results,the rolling force change,self-learning process,and strip elongation inconsistent with the length change after the control system is put into use are analyzed.The results show that the control system can accurately issue rolling process parameters,efficient self-learning optimization,and incremental learning,and successfully reduce the strip elongation mismatch length.Related control algorithms and control systems can provide help for the development of new intelligent leveling machine control systems.