Big data model: An application to design of rolling process

The current socioeconomic and climate trends imply that the sustainability of large industrial systems such as rolling industry must be urgently improved. Present design applications do not take sufficient advantage of big data accumulated in rolling mill repositories. At the same time, the nowadays information processors and analytical theories allow for performing real-time multivariate analysis and extracting important knowledge from industrial records. For this, however, there is a need to translate raw records into appropriate mathematical forms. The proposed approach to design of rolling process combines statistical analysis of rolling sequences with empirical and theoretical models of plastic flow. Deterministic models allow for creating rolling process that will function at some level of efficiency still below the possible higher level. This is evident due to the fact that the statistical analyses of production data recorded in identical mills show high dispersion. On the contrary to current methods, the proposed approach allows for diagnosing the causes for the existence of this gap, and also suggests how this gap can be decreased. An example of design of the leadeing oval groove for rolling wire rod is presented along with discussion of general mathematical aspects to demonstrate application of extracted statistics for probabilistic optimization of process parameters.