Primjena blisko infracrvene spektroskopije u analizi patvorenja meda

master's thesis

• Title (english): Application of near infrared spectroscopy in analysis of adulteration in honey
• Author: Lucija Longin
• Mentor: Ana Jurinjak Tušek (mentor)
• Granter: University of Zagreb; Faculty of Food Technology and Biotechnology; (Department of Process Engineering); (Laboratory for Measurement, Regulation and Automatisation); Zagreb
• Defense date and country: 2019-07-23, Croatia
• Scientific / art field, discipline and subdiscipline: BIOTECHNICAL SCIENCES; Food Technology

Abstract

Med je visokovrijedna namirnica, koja zbog svog sastava i cijene sve češće postaje meta patvorenja. U okviru ovog rada provedena je analiza fizikalno-kemijskih svojstava 15 uzoraka izvornih medova te patvorina meda s različitim udjelima komponenata za patvorenje (glukozni i kukuruzni sirup) te su razvijeni kemometrijski modeli (PLS i ANN) za opis i predikciju udjela dodane patvorine i fizikalno-kemijskih svojstva meda na temelju NIR spektara. Dobiveni rezultati fizikalno-kemijskih svojstva pokazuju da su analizirani uzorci izvornih medova u skladu s Pravilnikom o medu te da se dodatak patvorine jednostavno može detektirati na osnovu električne vodljivosti. Na temelju koeficijenta determinacije za validaciju, RMSEP vrijednosti, SEP i RPD vrijednosti za razvijene PLS modele za predikciju udjela patvorina u medu može se zaključiti da su razvijeni modeli robusni i pouzdani za primjenu na uzorcima koji nisu uključeni u model. Rezultati također upućuju da razvijene neuronske mreže osiguravaju dobro slaganje između eksperimentalnih vrijednosti i modelom previđenih vrijednosti na razini učenja, testiranja i validacije jer su za svih 15 uzoraka koeficijenti determinacije na sve tri razine (učenje, testiranje validacija) viši od 0,9 a pogreške modela niske (RMSE < 0,01) te se mogu koristiti za predikciju udjela patvorine i opis fizikalno-kemijskih svojstava patvorina meda na temelju NIR spektra.

Abstract (english)

Honey is a highly valuable food, and due to its composition and price is increasingly becoming the target of adulteration. In this work, the analysis of the physicochemical properties was made for 15 samples of the original honey, and adulterated honey with the different proportions of adulterants (glucose and corn syrup) and chemometric models (PLS and ANN) were developed for the description and prediction of the added adulterants and physicochemical properties of honey based on NIR spectra. The results of physicochemical properties indicate that the analyzed samples of original honey in complying with the Regulations about honey and that the addition of adulterants can easily be detected based on the properties of electrical conductivity. Based on the validation determination coefficient, RMSEP values, SEP and RPD values for developed PLS models for the prediction of adulterants ratio in honey, it can be concluded that the developed models are robust and reliable for use on samples that are not included in the model. The results also indicate that developed neuronal networks ensure a good correlation between experimental values and the model of predicted values at the learning, testing and validation level. For all 15 samples, the determination coefficients at all three levels (learning, validation testing) are higher than 0.9 and the model error is low (RMSE < 0.01) and can be used to predict the ratio of added adulterants and the description of the physicochemical properties of adulterated honey based on the NIR spectrum.

• Keywords: med; patvorenje; NIR spektroskopija; kemometrija
• Keywords (english): honey; adulteration; NIR spectroscopy; chemometrics
• Language: croatian
• URN:NBN: urn:nbn:hr:159:752849
• Study programme: Title: Food Safety Management; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka upravljanja sigurnošću hrane (magistar/magistra inženjer/inženjerka upravljanja sigurnošću hrane)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2019-12-05 12:39:48