Application of principal component-radial basis function neural networks （PC-RBFNN） for the detection of water-adulterated bayberry juice by near-infrared spectroscopy

Near-infrared （NIR） spectroscopy combined with chemometrics techniques was used to classify the pure bayberry juice and the one adulterated with 10% （w/w） and 20% （w/w） water. Principal component analysis （PCA） was applied to reduce the dimensions of spectral data, give information regarding a potential capability of separation of objects, and provide principal component （PC） scores for radial basis function neural networks （RBFNN）. RBFNN was used to detect bayberry juice adulterant. Multiplicative scatter correction （MSC） and standard normal variate （SNV） transformation were used to preprocess spectra. The results demonstrate that PC-RBFNN with optimum parameters can separate pure bayberry juice samples from water-adulterated bayberry at a recognition rate of 97.62%, but cannot clearly detect water levels in the adulterated bayberry juice. We conclude that NIR technology can be successfully applied to detect water-adulterated bayberry juice.

Application of principal component-radial basis function neural networks （PC-RBFNN） for the detection of water-adulterated bayberry juice by near-infrared spectroscopy

Near-infrared （NIR） spectroscopy, Principal component-radial basis function neural networks （PC-RBFNN）, Bayberry juice, Adulteration, Chemometrics technique