Investigating the process parameters that influence the z-strength of liquid paperboard using data mining and machine learning

Abstract: Parameters affecting the z-strength of liquid paperboard (LPB) has been analyzed and identifiedusing data mining and machine learning on 6 years of operational data from a multi-ply LPB mill, and control and stabilizing of them was proposed. Linear regression models were built for 9 articles with satisfactory results, and whose attributes were further analyzed as the most important parameters for the z-strength. The results show that generally only parameters affecting the weakest position in the paperboard has any influence on the z-strength, with unbleached softwood pulp refining work affecting the strength the most, while bleached hardwood refining work has a lower influence, and refining work of bleached softwood has almost no influence on the z-strength. Among the other parameters shown to influence the z-strength are kappa number, headbox concentration, broke ratio, strength and retention starches, fractionation and degree thereof, and the conductivity of the process water.