Simulation of Two Stands Cold Rolling Mill Process Using Neural Networks and Genetic Algorithms in Combination to Avoid the Chatter Phenomenon

Rolling mill Industry is one of the most profitable industries in the world. Chatter phenomenon is one of the key issues in this industry. Chatter or rolling unwanted vibrations not only has an adverse effect on product quality, but also reduces considerably the efficiency with reduced rolling velocities of rolling lines.This paper is an attempt to simulate the phenomenon of Chatter more accurate than the previous performed simulations. In order to increase the production speed, it needs to avoid parameters which effect on the Chatter and varieties with the rolling lines condition. Actual values of these parameters were determined in the archives of the Mobarakeh two stand cold rolling mills and collected on the 210 case study of real chattering. To simulate the experiment, a neural network is trained and weights and bias values of the neural network with genetic optimization algorithm were used to get an optimal neural network which reduces bugs on the test data. So this model is capable to predict speed of Chatter threshold on rolling process of two stand cold rolling mill with the accuracy less than one percent. So it can be used in rolling process with the building intelligent recognition systems to prevent the creator conditions of the chatter frequency range.