An innovative non-targeted control system based on NIR spectral information for detecting non-compliant batches of sweet almonds

摘要

Nowadays, there is growing awareness about the need to develop new methodologies to fight against deliberate fraud. This study explored the use of near infrared spectroscopy (NIRS) as an instantaneous, non-targeted method for detecting non-compliant products; in this case, when used to detect sweet almond batches adulterated with bitter almonds. For this purpose, we simulated the adulteration of batches by preparing four different types of mixed samples which contained 5%, 10%, 15% and 20% of bitter almonds, respectively, using 90 samples of sweet almonds and 50 samples of bitter almonds. For each of the adulteration percentages, 21 samples were produced. The samples were analysed using the Aurora and the Matrix-F spectrophotometers. The procedure initially constructed the desired standard or target using only the spectral information provided by the sweet almond population (control population). To achieve this, after principal components analysis, the spectral warning and action limits were calculated using the n-dimensional statistic Mahalanobis global distance. Next, the spectral distances from the product standard defined for those samples not belonging to the control population, including the adulterated sweet almonds, were calculated and represented as Shewhart control charts. The implementation of NIRS technology throughout the almond supply chain enabled to identify 87% (73/84) of the adulterated sweet almond batches. These findings suggest that NIRS technology and the use of spectral distances could enable to establish an innovative, non-targeted control system based only on spectral information to assess almond batches. This system allows to carry out conformity tests both in situ and online of the batches of almonds received and processed in the industry, as well as establishing fast, cost-efficient antifraud alert systems, which would help to reduce the number of batches to be analysed by expensive and time-consuming confirmatory methods. (C) 2020 Elsevier B.V. All rights reserved.