Study On The Application Of NIRS In Quality Analysis Of The Processing Adjunct Of Honey

Honey is one of the most used adjuncts for processing and producing Traditional Chinese Medicine (TCM). It has the function of improving taste and nutrition. The counterfeit honey products which mixed with glucose, sucrose and other sugars were often found in the markets. The conventional analytical methods of composition of honey, such as the method of optical rotation and colorimetric, are complex process and time-consuming. Therefore, it is in need of developing an effective method in identification of real or false honey and content determination. Near-infrared spectroscopy (NIRS) is a kind of rapidity and non-destructive analysis techniques which been successfully used in agriculture, food processing and oil industries. The thesis is to study the application of NIRS in authentic identification and content determination of the processing adjunct of honey.73 samples were collected in the study. The samples were assigned to calibration set and prediction set with the ratio of 4:1 by Kennard-Stone. Content of glucose and fructose in honey were determined by HPLC-ELSD, which are used as reference data. Multiple linear regression (MLR) was performed on the data, after the spectra were pretreated by mean centre, first derivative and orthogonal signal correction (OSC), to obtain calibration models. Correlation coefficient, standard error of calibration and standard error of prediction were applied to evaluate and compare each model's predictive ability. The MLR mathematical model that established with orthogonal signal correction plus first derivative and mean centre preprocessed spectra shows the best result for glucose and fructose. Correlative coefficient of glucose mathematical model was 0.912, with the standard error of calibration (SEC) of 1.21 and the standard error of prediction (SEP) of 0.94. Correlative coefficient of fructose mathematical model was 0.963, with SEC of 0.78 and SEP of 0.85. The study demonstrated that NIRS technology with MLR could be successful applied as a rapid and reliable method to determine the content of glucose and fructose in honey.In the aspect of identification of honey authenticity, the spectra were pretreated by mean centre and first derivative. Principle component analysis (PCA), Mahalanobis distance methods and Soft independent modeling of class analogy (SIMCA) was used to establish the identification model. The result showed that the established model could identify the honey whether it was real or false accurately but not the type of honey.The conclusion of the study was that NIRS technology could be applied to identify of honey authenticity and determine the content of glucose and fructose in honey effectively.