In this paper, we present the application of data mining techniques to develop an "artificial lock" for the skin-pass in an attempt to solve a problem that can arise during the galvanising manufacturing process: the wrong labelling of the steel grade of a coil. In order to detect these errors and thus to avoid that coils with different properties than expected end up with a client, we propose neural network-based models for on-line predicting the strip elongation in the skin-pass section according to the manufacturing conditions and its chemical composition. Thus, a significant difference between estimated and measured elongation would mean that the coil must be removed from the line for further analyses.
González-Marcos, A., Ordieres-Meré, J. B., Pernía-Espinoza, A. V., & Torre-Suárez, V. (2008). Development of an artificial lock for the skin-pass section in a hot dip. Revista de Metalurgia, 44(1). https://doi.org/10.3989/revmetalm.2008.v44.i1.93