Prediction And Self-learning Of Rolling Force For The Aluminum Tandem Hot Rolling Mill

With the improving quality requirements on the market, aluminum hot rolling controlsystem is becoming a hot spot of research. However, level two control system foraluminum hot rolling has not been developed independently in China. The domestic leveltwo control system for aluminum hot rolling is imported from abroad with a high price. Sothe development of the level two control system must be speeded up.Under the background of basic automation devices transformation and level twocontrol system optimization project for2000mm aluminum tandem hot rolling, basicautomation devices transformation has been completed, and level2control system will beupgrade next.The setting calculation of the rolling model is a key part of the level twocontrol system. And rolling force model is the most important model of all rollingmathematical models.Based on lots of actual rolling data, the value of deformationresistance was calculated through rolling force model. At the same time, least squaresmethod was used to regress the coeffiient of the deformation resistance model of differentalloy. According to lots of actual rolling data, the actual deformation resistance model wasestablished by the method of Least Squares Support Vector Machine LSSVM. BacteriaForaging Optimization (BFO) algorithm was applied to optimize the gain coefficient ofthe model and the accuracy of the model was improved.Aluminum strip rolling process is very complex. Many factors cause rolling forceprediction error, such as the error of the model, the measurement error, the change of theprocess state. In order to improve the prediction precision of rolling force, theself-learning method was applied to rolling force model. This theory was applied to thesimulation experiment of the rack4and result shows that prediction precision of rollingforce has been effectively improved to10%.