Stretch-reducing mill : a new approach by neural networks

Abstract

A neural network model, trained by a genetic algorithm, has been developed to study the thickened ends of the pipes rolled by means of two stretch-reducing mills. The forecasting of the model showed a good performance that avoids other complicated developments based on precise structural approaches that generally take a great amount of elaboration time. The neural network can be applied to build an automatic procedure for the definition of the production parameters to minimize the discarded material of the pipe that is not in the limits imposed by the geometrical specifications.