Roller Thermal Deformation Prediction and Constant Gap Control in Planar Flow Casting

Abstract

The effects of nozzle length and width, alloy injecting speed, as well as roller thickness and speed on roller thermal deformation were investigated by thermal-structure coupling method based on the non-uniform heat flow boundary. Furthermore, a neural network prediction model of roller thermal deformation based on generalized regression neural networks (GRNN) was constructed, and a new control strategy of the constant gap based on the prediction model was proposed. The results show that the influence of nozzle width and length on roller deformation is more obvious than that of roller speed and thickness; the accuracy of the proposed neural network prediction model based on GRNN is acceptable, and the average relative error is 5.98%; compared with the traditional PID control method, the proposed control strategy of the gap has no lag and can meet the requirements of fast and accurate tracking of cooling roller deformation, ensuring a constant gap and significantly improving the qualification rate and production efficiency during the initial preparation of thin ribbons.

Keywords

• planar flow casting
• cooling roller
• thermal deformation
• neural network prediction
• constant gap control