Rapid Identification of Herbaceous Biomass Based on Raman Spectrum Analysis

Abstract

To realize the rapid online identification of tobacco quality style in a few seconds, this paper proposed a rapid identification method of tobacco quality style based on the Raman spectrum analysis. This method could quickly obtain the Raman spectrum of tobacco samples, then establish the mapping relationship database between the tobacco information and the Raman signal. The KNN (K-Nearest Neighbors) algorithm as the identification algorithm of tobacco information was used. The rapid and accurate identification of tobacco origin through the Raman signal in the range of 2700–3500 cm−1 was realized. The accuracy of origin identification can reach 95.7%. To identify the tobacco grade accurately, the competitive adaptive weighted sampling (CARS) was used to select the key Raman characteristic spectral coverage that determines the acteristics of tobacco grade. Combined with the KNN algorithm, the rapid and accurate identification of tobacco grades by analyzing the signals of key Raman characteristic spectral coverage was realized. By using the key Raman characteristic spectral coverage in the range of 800–800 cm−1 and 2700–3500 cm−1, the accuracy of tobacco grade identification can reach 87.0%. The results showed that by analyzing the key Raman characteristic band signals of unknown tobacco samples, combined with the identification algorithm proposed in this paper, the efficient identification of the quality style of unknown tobacco can be realized.