Evaluation of in-mold sensors and machine data towards enhancing product quality and process monitoring via Industry 4.0

Abstract

With the rise of Industry 4.0-related technology in the plastic and composite industry, a new wealth of data from the production process is becoming available to manufacturers. The effective utilization of this data towards improving quality and output is therefore of critical importance but requires knowledge of the data that is truly useful and the application of that data to pre-developed models or trained algorithms. Accordingly, in this research, 12 different online data sources in the injection molding process are evaluated to determine their relative degree of importance in predicting variations on final part quality indices, namely part weight, thickness, and diameter. These data are obtained during each injection molding cycle using a data acquisition system connected to eight in-mold sensors and four machine data sources. Three distinct types of perturbations are introduced into the process to challenge the range of detection capacities of these various data sources: shot size variations, material disturbances, and shutdown of the mold cooling system. The resultant curves from these studies are then analyzed for critical values, and partial least square (PLS) regressions performed using the extracted values as predictors and the final part quality indices as responses. Using the standard coefficients from the PLS analysis, rankings of the correlations between the extracted values and final part quality indices are generated, indicating which data sources best detected variations in the final produced parts for each of the three perturbations.