Neural network approach to quality monitoring of injection molding of photoluminescent polymers

Photoluminescent polypropylene was obtained by twin-screw extrusion, the engineered material was injection molded and the resulting flat components were characterized in terms of luminescence and mechanical properties. Different pigment concentrations and different thickness of molded samples were considered. Instrumented flat indentations were performed, stating good mechanical performance (more than 500 N indentation load at 0.3 mm penetration depth) of components, which showed at the same time a reliable photoluminescent emission over a reasonable time range (persistence time 30 min). The experimental trend of load-penetration curves was modelled by an artificial neural network. A good generalization capability and high flexibility were found for the proposed neural network solution.