Detection of Damaged Tool in Injection Molding Process with Acoustic Emission

ABSTRACT

The measurement of acoustic emission (AE) signals during injection molding of polypropylene with new and damaged mold is presented. The damaged injection mold was fitted with a steel insert with cracks induced by laser surface heat treatment. Two resonant piezoelectric AE sensors were attached to the mold via AE waveguides. To improve the mold integrity prediction with smaller defects, AE signal frequency characteristics and a measure of AE signal amplitude probability distribution are implemented. A 5-dimensional feature vector with real-valued explanatory variables is proposed, providing the defining points in an appropriate multidimensional space to characterize the state of injection molding tool. Feature vectors are classified with neural network pattern recognition. The results confirm that presented AE technique offers characterizing the integrity of molds also with resonant sensors.