Using The Finite Element Method And Artificial Neural Networks To Predict Strip Width In Hot Strip Rolling

摘要

In hot strip rolling the strip width is determinedby the initial slab width and the rolling procedure in theroughing and finishing rolling mills. The width can to someextent be controlled by edge rolling in the roughing mill. TheFinite Element Method (FEM) has proved to be a powerful toolto investigate three-dimensional material flow in rolling.Below is described how FEM is used to calculate the materialbehaviour in the roughing stand of a hot strip rolling mill. Hotcompression tests are done for different temperatures,deformations and deformation rates to get data for calibrationof the material model that is used. Of special interest are thedeformations occurring at the strip edges due to alternatinghorizontal and vertical rolling. Formation of so called"dogbone" and "edge bulge" profiles is considered. Thecalculation results have been compared to measurements withgood agreement. Experimental design is used to generate FEMcalculation cases with independent variation of all importantrolling parameters. Results from the FEM calculations are usedto train Artificial Neural Networks (ANN' s). ANN predictionsare compared with FEM calculations and measurements on rolled strips.