Detection Of Honey Quality Based On Near Infrared Spectroscopy

With the improvement of people’s quality of life and the strengthening of health awareness,honey as a natural agricultural product is favored by consumers.Due to the chaos in the product market of bee,the classification of honey varieties and the adulteration of the honey have a great impact on the quality of honey.The traditional honey quality identification is mainly based on the combination of sensory evaluation and physical and chemical analysis,which is time-consuming and labor-intensive.Therefore,it is a great significance to study a method for rapid honey quality identification.The research content of this paper includes the following two parts.Firstly,three honeys of acacia honey,red jujube honey and rapeseed honey are taken as experimental objects,and their spectral reflection images are obtained by near-infrared spectrum image acquisition system.Spectral features were extracted from the pre-processed spectral reflectance curves,and a honey species classification model based on nearinfrared spectroscopy was established.(1)The full spectral characteristics combined with BP and SVM can accurately classify honey varieties of different varieties to 89.58%.(2)In order to screen the feature variables and eliminate data redundancy,the principal component PCA feature extraction and SPA feature selection methods are used respectively,and the selected features are combined with BP neural network and SVM support vector machine classification model.After analysis,PCA-SVM is the optimal honey species classification model,and the correct classification rate of honey varieties can reach 93.75%.After determining the spectral curve of the near-infrared spectrum image,the honey species can be classified,and the possibility of judging the pure honey after adulteration is further explored.Using the spectral reflectance curve to establish a honey adulteration identification model based on partial least squares discrimination PLS-DA,the classification accuracy was 97.92% in the test set.The research results show that the PLS-DA model has high accuracy to identify honey adulteration ability.