Modelling of hot strip rolling process using a hybrid neural network approach

A neural network-based approach is developed to predict a mechanical property for the hot-rolled alloy strip. Using a data set containing critical information on the mechanical property which was obtained from a POSCO hot strip mill, a neural network-based model is elicited. A compact set of process variables is selected as the model inputs, based on expert's knowledge as well as a correlation analysis technique. The ensemble modelling technique is used to improve the model performance and to provide error bounds for the prediction error very close to the maximal measurement standard deviation. It is concluded that the developed neural network model is capable of providing good prediction accuracy with a very fast computing speed and some degree of transparent interpretability of the normally opaque structure of neural networks.