Study On Rapid Identification Technology Of Honey Varieties

Honey is widely popular with its rich nutritional value,but there are adulteration and mixed varieties of honey in the market,which seriously harm the interests in consumers.Therefore,it is important to detect the quality of honey and identify honey varieties using rapid and accurate quality detection technology.In order to quickly identify the quality of honey,the physical properties of honey,such as,optical properties,texture properties and thermal properties were studied.The main research contents and results were as follows:1.The optical properties of different honey varieties were studied by NIR spectroscopy.The difference of NIR spectrum of 22 ℃ commercially available honey was the largest,and the spectral information was preprocessed and feature extraction was carried out.The absorbance at two bands of 840～910 nm and 940～980 nm was selected as the eigenvalues,and the eigenvalues were normalized and modeled by Fisher discriminant analysis method.The classification accuracy of the model is 85.5%,and the validation accuracy is 76.2%.2.The optical properties of different kinds of honey were studied by UV-VIS spectroscopy.The UV-VIS spectrum of the sample with 97% water content at 22 ℃ showed the greatest difference,and the spectral information was preprocessed and feature extracted.The absorbance at four wavelengths of 225 nm,260 nm,300 nm and 320 nm was taken as the eigenvalues,and the eigenvalues were normalized and modeled by Fisher discriminant analysis method.The classification accuracy of the model is 98.6%,and the validation accuracy is 95.2%.3.Texture properties of different varieties of honey were studied by texture properties determination.The determination of honey by texture apparatus conforms to the principle of fluid flow in circular pipes.At 22 ℃,the texture data of the three kinds of commercially available honey showed the greatest difference.The data at 0s,5 s,10 s,15 s,20 s and 25 s were selected as the eigenvalues,and the eigenvalues were normalized and modeled by Fisher discriminant analysis method.The classification accuracy of the model is 63.8%,and the validation accuracy is 61.9%.4.The thermal properties of different honey varieties were studied by differential scanning calorimetry.With the increase of water content,the glass transition temperature,initial phase transition temperature,peak phase transition temperature and end phase transition temperature of honey decreased,while the melting temperature and enthalpy value increased.Commercially available honey and samples with water content of 30% showed glass transition,while samples with water content of 40% and 50% did not show glass transition,but appeared melting phenomenon.The glass transition temperature of the three kinds of commercially available honey was significantly different.However,due to the small sample size,whether it can be used to identify honey varieties needs further study.5.The data fusion was used to merge the NIR,UV-VIS and texture information of honey after pretreatment and feature extraction.After normalization,the identification model of honey varieties was established.The classification accuracy of the model is 100%,and the validation accuracy is 100%.According to the above results,differential scanning calorimetry is not used for data fusion modeling due to its lack of regularity and small sample size.NIR spectroscopy,UVVIS spectroscopy and texture property determination can be used to identify honey varieties,but the classification ability of honey variety identification model based on data fusion is better.