engleski

Application of artificial neural networks coupled to NIR-spectroscopy for detection of adulterations in honey

Honey is a naturally sweet and viscous product for which the addition of any substance is prohibited by international regulation. The detection of the adulterations in the honey is technical problem, due to the adulteration of honey with invert sugar and syrup may not be reliably detected by direct sugar analysis because its constituents are the major natural components of the honey. Therefor it is import to develop a rapid and reliable analytical method. In this work near-infrared spectroscopy (NIRs) combined with principle component analysis (PCA) and artificial neural networks modelling (ANN) was used for discrimination of honey adulterated with corn syrup. Fifteen honey samples from the north-west part of Croatia (Krapina-Zagorje County) were adulterated with corn syrup in the range from 10-90 %. Totally 460 spectra were collected by NIR spectrophotometer NIR128L-1.7 (Control Development, South Bend, Indiana, USA) with installed Control Development software Spec32 using a halogen light source (HL-2000) for the wave length range of  = 904 – 1699 nm. For each of the prepared sample, water content was measured by refractometer, conductivity by conductometer (SevenCompact, MettlerToledo, Switzerland) and colour by colorimeter PCE-CSM3 (PCE Instruments, Germany). Prior to ANN modelling, PCA was used for identifying patterns and to highlight similarities and differences in data of the individual set of the experiments. The goal of PCA is to extract the important information from the data table and to express this information as a set of new orthogonal variables called principal components or factors (PCs or Fs). Raw spectra were used to perform PCA by the Unscrambler® X 10.4, software (CAMO software, Norway). Based on gathered NIR spectra, ANN calibration models were developed for simultaneous prediction of amount of added adulteration in the honey, water content, conductivity and colour of the adulterated honey. Multiple layer perceptron (MLP) networks were developed in Statistica v.10.0 software (StatSoft, Tulsa, USA). The artificial neural network (ANN) training was performed with separation of data into training, test and validation sets as 70:15:15 ratio using first five factors form the PCA analysis as the input variables. Back error propagation algorithm available in Statistica v.10.0 was applied for the model training. The model performance was evaluated based on R2 and Root Mean Squared Error (RMSE) values for training, test and validation. Obtained results show that network MLP 5-8-6 with five neurons in input layer, 8 neurons in hidden layer and 6 neurons in output layer predicts the analysed output variables with high precision (R2validation, concentration = 0.995, R2validation, water content = 0.993, R2validation, conductivity = 0.992, R2validation, L = 0.939, R2validation, a = 0.895, R2validation, b = 0.924).

Near-infrared spectroscopy (NIR), principle component analysis (PCA), artificial neural networks modelling (ANN), honey adulteration

nije evidentirano