Premium
Artificial neural network approach to predict the lightfastness of gravure prints on the plastic film
Author(s) -
Mandal Mahasweta,
Bandyopadhyay Swati
Publication year - 2020
Publication title -
color research and application
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.393
H-Index - 62
eISSN - 1520-6378
pISSN - 0361-2317
DOI - 10.1002/col.22504
Subject(s) - materials science , spectroradiometer , artificial neural network , computer science , environmental science , statistics , reflectivity , optics , mathematics , artificial intelligence , physics
The lightfastness of prints is an important property for assessing their print stability. The objective of this study is to determine the lightfastness rate of printed films due to long‐time exposure by applying artificial neural network (ANN). Package printing is gradually becoming extremely important because its color and quality increase the marketability of the product. Sometimes it has been observed that the initial print quality is bright and attractive. However, with time, it degrades the exposure of light, water, or other external parameters. Thus, it reduces the marketability if its color degrades before its expiration. Therefore, the lightfastness of prints may be considered for the authenticity or validity of the product. The plastic film substrate is chosen because it has extensive usage in food and other packaging industries. The samples printed in the gravure process are exposed in artificial lightfastness tester BGD 865/A Bench Xenon Test Chamber (B‐SUN) for assessing the lightfastness of prints. The ocean optics spectroradiometer (DH2000BAL) is used to measure the spectrophotometric properties of prints before and after exposure. The obtained reflectance spectra are modeled by applying an ANN technique that is proposed to predict the fading rate of the printed film. The optimal model gives excellent prediction with the minimum mean square error for each color and a correlation coefficient of 0.80 to 0.99. ANN model and a Regression model (assuming first‐order kinetic equation) are compared for predicting the lightfastness properties of prints. The results show that the ANN has better prediction capability than the regression model.