Prediction of reversible cold rolling process parameters with artificial neural network and regression models for industrial applications: A case study

摘要

Reversible cold rolling process is a well-known method of metal forming. Since the process is highly-automated, predicting process parameters is essential to optimize processing time. An iterative algorithm has been developed and integrated into the industrial reversible cold rolling process. Regression and artificial neural network algorithms have been used and compared for prediction. To obtain high accuracy, the best fitted algorithms have been selected in each pass. Moreover, regression-ANN hybrid algorithms have been developed for the cases where one algorithm is insufficient to predict all parameters accurately. Finally, processing times have been calculated for the optimization process.