Neural networks and fuzzy rules based control for cold rolling process via sensitivity factors

A method for the calculation of the appropriate adjustment of the three control parameters (roll gap, front or back tensions) and an application of neural control to rolling mill are presented. The method uses the sensitivity equation of the process and fuzzy rules, obtained by differentiating a neural network. The method to obtain "fuzzy rules" of physical processes, is a new technique to extract knowledge of the same ones, without need to obtain complex analytic expressions based on models. This method based in the sensitivity factors of the process can contribute to the development of a new technology utilized in online supervision and control systems, where the computational efforts gets to be critical