Nondestructive Quality Evaluation of Intact Mangos Using Near Infrared (NIR) Spectra Combined by Artificial Neural Networks and Multivariate Analysis
Quality attributes of agricultural products encompasses sensory properties, nutritive values, chemical constituents, mechanical properties, functional properties and defects which are determined normally by wet chemical analysis in which destruct the object. It also took some time and involves chemical materials. Based on this point of view, non-destructive techniques for evaluating quality parameters of different agricultural commodities are gaining attention. The advantages of these techniques include rapid execution, no chemical waste and without destruct the object. Near infrared reflectance (NIR) is one of non-destructive technique that works based on the principle of interaction of electromagnetic radiation with biological matters. Thus, the objectives of this study are: (1) to detect and classify mango cultivars and origins based on near infrared spectra signatures using Principal Component Analysis and Probability Neural Networks by developing artificial neural networks algorithm that can be distinguish mango cultivars and origin based on NIR spectral data. (2) To determine inner quality attributes of mango (total soluble solids, ascorbic acids and total acidity) based on NIR spectra using Principal Component, Partial Least Squares Regression (PCR and PLSR) and Generalized Regression Neural Networks by developing and quantifying calibration models.
Fig. 1: Modelling of the qualitiy attributes based on NIR spectra
Am Projekt beteiligte Mitarbeiter: