Data-driven models for a salt production process towards an Industry 4.0 evolution

This work develops a predictive model of the production process of brine salt in an Italian industrial site. The methodology uses dimensionality reduction via standard statistical techniques and one year of production data has been acquired via direct connection to the plant control system. A code developed in Python analyze each plant, screen the raw data, and regress the models via principal component regression (PCR) and partial least squares (PLS). Results show good reliability for the prediction of the evaporative plant while the depuration model still needs refinements to be performed.