Hybrid neural–GA model to predict and minimise flatness value of hot rolled strips

Abstract The profile and flatness of a strip are key attributes in determining the strip quality. Thinner gauge hot rolled (HR) strips have been found more prone to flatness defects as compared to thicker gauge HR strips. In order to achieve a good profile and flatness it is required to select the most optimum process parameters. In this work an attempt has been made to minimise the flatness defects in HR strips using a novel hybrid model with the combination of a predictive artificial neural network (ANN) and a genetic algorithm (GA). GA is meant for the optimisation of few selected process parameters affecting the flatness of the strip and forms a closed loop with the ANN. The predictive model has been tested and the flatness values are minimised for a set of selected strips.