Classification of olive oils according to sensory defects using a potentiometric electronic tongue
Conference Paper
Overview
Research
Additional Document Info
View All
Overview
abstract
Olive oil is a highly appreciated food product being very prone to frauds. Olive oils may be
graded as extra-virgin, virgin or lampante. This classification is attributed according to legal
requirements, including chemical parameters and sensorial analysis. Among the
organoleptic sensations, the capability of perceiving the presence or absence of sensory
defects plays a key role for olive oils grade classification. This task is time-consuming and
quite expensive, requiring the use of an official taste panel, which can only evaluate a low
number of samples per day. In this work, an electronic tongue is proposed to discriminate
olive oils according to the defect predominantly perceived (winey-vinegary, wet-wood,
rancid and fusty/muddy sediment), by a trained sensory panel. Sub-sets of potentiometric
signal profiles obtained from the lipid sensor membranes of the taste electrochemical
device were selected using a simulated annealing meta-heuristic algorithm, allowing
establishing classification linear discriminant model, which showed a predictive success
classification rate of 81% for leave-one-out or cross-validation procedure. The satisfactory
predictive performance achieved pointed out the practical potential of using this artificial
taste sensor as a complementary methodology for olive oil sensory analysis.
Olive oil is a highly appreciated food product being very prone to frauds. Olive oils may be graded as extra-virgin, virgin or lampante. This classification is attributed according to legal requirements, including chemical parameters and sensorial analysis. Among the organoleptic sensations, the capability of perceiving the presence or absence of sensory defects plays a key role for olive oils grade classification. This task is time-consuming and quite expensive, requiring the use of an official taste panel, which can only evaluate a low number of samples per day. In this work, an electronic tongue is proposed to discriminate olive oils according to the defect predominantly perceived (winey-vinegary, wet-wood, rancid and fusty/muddy sediment), by a trained sensory panel. Sub-sets of potentiometric signal profiles obtained from the lipid sensor membranes of the taste electrochemical device were selected using a simulated annealing meta-heuristic algorithm, allowing establishing classification linear discriminant model, which showed a predictive success classification rate of 81% for leave-one-out or cross-validation procedure. The satisfactory predictive performance achieved pointed out the practical potential of using this artificial taste sensor as a complementary methodology for olive oil sensory analysis.
Olive oils may be graded according to their overall
physicochemical composition and sensorial attributes as:
-extra-virgin olive oils (EVOOs);
-virgin olive oils (VOOs);
- lampante olive oils (LOOs).
Olive oils are quite prone to frauds thus there are legal
protection EU Commission regulations:
- EU Commission Regulation, 1991;
- EU Commission Regulation, 2011.
Maximum levels are established for:
- Chemical and physicochemical parameters (e.g., free
acidity, peroxide value, UV extinction coefficients and
alkyl esters content)i
-Sensory attributes (presence/absence of organoleptic
defects, fruity sensation and positive attribute).
This work was fiancialy supported by Project POCI-01-0145-FEDER-006984 - Associate Laboratory LSRE-LCM, Project UID/QUI/00616/2013 - CQ-VR, Project UID/BIO/04469/2013 - CEB and Project UID/AGR/00690/2013- CIMO all funded by FEDER · Fundo Europeu de Desenvolvimento Regional through COMPETE2020 - Programs Operacional Competitividade e lnternacionalização (POCI) - and by national funds through FCT - Fundação para a Ciênciaa e a Tecnologia, Portugal. Nuno Rodrigues thanks FCT. POPH-QREN andFSE for the Ph.D. Grant SFRH/BD/104038/2014
This work was financially supported by Project POCI-
01–0145-FEDER-006984 – Associate Laboratory
LSRE-LCM and by Project UID/QUI/00616/2013 –
CQ-VR both funded by FEDER - Fundo Europeu de
Desenvolvimento Regional through COMPETE2020 - Programa Operacional Competitividade e
Internacionalização (POCI) – and by national funds
through FCT - Fundac ão para a Cie ncia e a
Tecnologia, Portugal. Strategic funding of
UID/BIO/04469/2013 unit is also acknowledged.
Nuno Rodrigues thanks FCT, POPH-QREN and FSE
for the Ph.D. Grant (SFRH/BD/104038/2014).
