Modelization of ironing process. Application to a three-layered polymer coated steel

Abstract

This document shows a modelization for the ironing process (the most crucial step in can manufacturing) done by using a Neural Network. A design of experiments has been done, in which a number of process variables have been taken in consideration, controlled directly by the operator: die angle, punch speed, temperature and reduction in thickness. These will be the process inputs, whereas the exits are formed by two variables result: Surface quality factor and average roughness. Once the results have been obtained, data have been introduced in a neural network software, so that after being trained the network and used the BP algorithm (BackPropagation), generates a configuration that provides the smaller permissible error. Later, by means of the sensitivity analysis, the neural network generated has been simplified with a new configuration of a minimum error.