Study On Carrot Juice Identification Based On Raman Spectroscopy And SVM

In recent years,food safety incidents have caused serious harm,and food detection has become increasingly important.Traditional food detection technologies have their own limitations,while Raman spectroscopy has the advantages of fast,non-destructive,high detection sensitivity and wide detection range.The Support Vector Machine(SVM)algorithm proposed for small samples,high dimensionality and nonlinear problems has been widely used in the field of pattern recognition.Some scholars have combined Raman spectroscopy and SVM to achieve qualitative analysis of food,but there is currently no good method for SVM parameter selection in the analysis process.In summary,in order to identify the types of carrot juice quickly and non-destructively,this paper used SVM to identify the Raman spectrum of carrot juice,the main research work is as follows:In order to improve the classification performance of SVM,an in-depth study on the selection of its penalty factor and kernel function parameters was carried out,and the Particle Swarm Optimization algorithm(PSO)was used to optimize the SVM parameters to construct the classifier pso SVM.In order to make up for the defects of this classifier,the Simulated Annealing algorithm(SA)was used to improve it,and the classifier sapso SVM was constructed and its classification performance was verified on the data set.The research results show that the sapso SVM classifier proposed in this paper has better classification performance than pso SVM.This work not only provides a new method for SVM parameter optimization,but also lays a good foundation for identifying Raman spectrum signal.It was further proposed that the combination of Raman spectroscopy and pattern recognition algorithms be applied in the identification of carrot juice types.First of all,the Raman spectrum signal of carrot juice was smoothed and denoised by Savitzky-Golay Filter(SGF),and the baseline was corrected by Adaptive Iteratively Reweighted Penalized Least Squares(AIRPLS),and then the data was normalized;Secondly,Principal Component Analysis(PCA)and Generalized Discriminant Analysis(GDA)were used to reduce the dimension of the data;Finally,K-Nearest Neighbor(KNN),Back Propagation Neural Network(BPNN),pso SVM and sapso SVM were used to classify the data after dimensionality reduction to establish 8classification models and comparatively analyze their classification performance.The results verify the feasibility of Raman spectroscopy combined with pattern recognition algorithms in the identification of carrot juice types.The results show that the classifier sapso SVM proposed in this paper has better classification performance than pso SVM.The classification model GDA-sapso SVM can effectively identify carrot juice Raman spectrum with a classification accuracy of92.38%.The study provides a efficient and convenient new method to identify carrot juice types.