摘要
In this study, a seed origin discrimination model for Clinacanthus nutans was developed. First, 81 C. nutans samples from three seed origin locations were collected, and their Near-Infrared (NIR) spectra were obtained. Next, Principal Component Analysis (PCA) was performed on the NIR spectra of the 81 C. nutans samples. Then, MSC (multiplicative scatter correction), SNV (standard normal variate), first derivative, and second derivative pre-treatments of the C. nutans spectra were performed and combined with the Support Vector Machine (SVM) algorithm for modelling and analysis. Among these methods, first-order derivative pre-treatment achieved the best SVM model effectiveness, with a training set accuracy of 93.44% (57/61) and a test set accuracy of 85.00% (17/20). In order to further improve the discrimination accuracy of the model, three optimization algorithms Grid Search (GS), Genetic Algorithm (GA), and Particle Swarm Optimization (PSO) were employed to identify the best c and g parameters for the SVM model. The results demonstrated that the PSO optimization algorithm yielded the best parameters of c = 0.8343, g = 57.8741, with corresponding model training set the accuracy of 96.36% (60/61) and test set the accuracy of 95.00% (20/21). Therefore, developing a seed origin classification model for C. nutans based on NIR spectroscopy combined with chemometrics is feasible and has the advantages of being simple, rapid, and green.
In this study, a seed origin discrimination model for Clinacanthus nutans was developed. First, 81 C. nutans samples from three seed origin locations were collected, and their Near-Infrared (NIR) spectra were obtained. Next, Principal Component Analysis (PCA) was performed on the NIR spectra of the 81 C. nutans samples. Then, MSC (multiplicative scatter correction), SNV (standard normal variate), first derivative, and second derivative pre-treatments of the C. nutans spectra were performed and combined with the Support Vector Machine (SVM) algorithm for modelling and analysis. Among these methods, first-order derivative pre-treatment achieved the best SVM model effectiveness, with a training set accuracy of 93.44% (57/61) and a test set accuracy of 85.00% (17/20). In order to further improve the discrimination accuracy of the model, three optimization algorithms Grid Search (GS), Genetic Algorithm (GA), and Particle Swarm Optimization (PSO) were employed to identify the best c and g parameters for the SVM model. The results demonstrated that the PSO optimization algorithm yielded the best parameters of c = 0.8343, g = 57.8741, with corresponding model training set the accuracy of 96.36% (60/61) and test set the accuracy of 95.00% (20/21). Therefore, developing a seed origin classification model for C. nutans based on NIR spectroscopy combined with chemometrics is feasible and has the advantages of being simple, rapid, and green.
作者
Fengyan Yu
Jinfang Ma
Yi Qi
Han Song
Guiliang Tan
Furong Huang
Maoxun Yang
Fengyan Yu;Jinfang Ma;Yi Qi;Han Song;Guiliang Tan;Furong Huang;Maoxun Yang(Guangdong Provincial Key Laboratory of Research and Development of Natural Drugs, the Second Clinical Medical College, School of Pharmacy, Guangdong Medical University, Dongguan, China;Opto-Electronic Department of Jinan University, Guangzhou, China;The Marine Biomedical Research Institute of Guangdong Zhanjiang, Zhanjiang, China;Marine Chinese Medicine Branch, National Engineering Research Center for Modernization of Traditional Chinese Medicine, Zhanjiang, China;School of Material Science and Food Engineering, University of Electronic Science and Technology of China, Zhongshan Institute, Zhongshan, China)
