Spectroscopy and Spectral Analysis, Volume. 42, Issue 6, 1774(2022)
Research on Identification of Danshen Origin Based on Micro-Focused Raman Spectroscopy Technology
Figures & Tables(9)
Fig. 1. Optimization results of random forest model parameters
(a): The relationship diagram between n_estimator and OOB_SCORE; (b): The relationship diagram between max_features and OOB_SCORE
Fig. 2. The relationship between the importance of spectral variables after 1st-d processing and wave numbers
Fig. 2. Original spectra of surface of danshen with different scanning times and tanshinone components
Fig. 3. The confusion matrix of the test set of the optimal RF model (a) and the optimal RF-VS model (b)
Table 1. Information of 150 samples of danshen
View tableView in ArticleTable 1. Information of 150 samples of danshen
采样区域
(见图1 )采样点
(见图1 )样本数量 土壤类型 地貌 a 粉色点 17 沙质 平原 b 深蓝色点 17 沙质 平原 c 亮绿色点 67 黄棕 丘陵 d 红色点 30 黄棕 丘陵 e 黄色点 19 黄棕 丘陵
Table 1. The results of best models under the conditions of different scanning times
View tableView in ArticleTable 1. The results of best models under the conditions of different scanning times
扫描次数 模型 数据前处
理方法训练集 测试集 变量数
/个界值 ACC ACC 扫描一次 RF-VS 1ST-D 88 87 167 0.009 207 扫描二次 RF-VS 3RD-D 89 87 99 0.013 165 扫描三次 RF-VS 3RD-D 89 87 307 0.008 811
Table 2. The results of the model under the conditions of different determination times
View tableView in ArticleTable 2. The results of the model under the conditions of different determination times
测定次数 模型 数据前处理方法 训练集 测试集 主成分数 变量数/个 界值 ACC ACC 测定一次 PLS-DA Original data 73 68 2 3 216 MSC 74 65 3 3 216 SNV 69 68 2 3 216 1ST-D 74 68 2 3 215 2ND-D 69 68 2 3 214 3RD-D 69 65 2 3 213 RF Original data 83 87 3 216 0 MSC 86 87 3 216 0 SNV 82 68 3 216 0 1ST-D 87 81 3 215 0 2ND-D 87 84 3 214 0 3RD-D 87 81 3 213 0 RF-VS Original data 86 87 87 0.013 837 MSC 86 87 366 0.008 228 SNV 81 77 295 0.010 197 1ST-D 88 87 167 0.009 207 2ND-D 88 81 29 0.014 741 3RD-D 88 87 198 0.009 325 测定二次 PLS-DA Original data 71 68 2 3 216 MSC 74 65 3 3 216 SNV 70 65 2 3 216 1ST-D 73 68 2 3 215 2ND-D 70 68 2 3 214 3RD-D 69 65 2 3 213 RF Original data 87 81 3 216 0 MSC 85 84 3 216 0 SNV 82 71 3 216 0 1ST-D 88 77 3 215 0 2ND-D 88 77 3 214 0 3RD-D 88 77 3 213 0 RF-VS Original data 86 84 3 215 0.011 960 MSC 87 84 101 0.007 782 SNV 84 77 319 0.008 426 1ST-D 87 84 3 214 0.009 207 2ND-D 87 87 3 213 0.014 382 3RD-D 89 87 99 0.013 165 测定三次 PLS-DA Original data 71 68 2 3 216 MSC 74 65 4 3 216 SNV 71 65 2 3 216 1ST-D 73 68 2 3 215 2ND-D 69 68 2 3 214 3RD-D 68 68 3 3 213 RF Original data 85 84 3 216 0 MSC 82 87 3 216 0 SNV 84 74 3 216 0 1ST-D 89 84 3 215 0 2ND-D 89 81 3 214 0 3RD-D 90 77 3 213 0 RF-VS Original data 86 87 78 0.011 553 MSC 85 84 156 0.010 343 SNV 83 77 218 0.008 102 1ST-D 89 84 303 0.009 168 2ND-D 89 84 167 0.017 988 3RD-D 89 87 307 0.008 811
Table 3. Test set confusion matrix of PLS-DA model after MSC processing
View tableView in ArticleTable 3. Test set confusion matrix of PLS-DA model after MSC processing
Memerbers ACC/% a b c d e 4 0 a 0 0 4 0 0 2 100 b 0 2 0 0 0 12 83 c 0 0 10 2 0 9 100 d 0 0 0 9 0 4 0 e 0 2 2 0 0 0 No class 0 0 0 0 0 1 67.74 Total 0 4 16 11 0
Tools
Get Citation
Copy Citation Text
Qing LI, Li XU, Shan-gui PENG, Xiao LUO, Rong-qin ZHANG, Zhu-yun YAN, Yong-sheng WEN. Research on Identification of Danshen Origin Based on Micro-Focused Raman Spectroscopy Technology[J]. Spectroscopy and Spectral Analysis, 2022, 42(6): 1774
Paper Information
Category: Research Articles
Received: Mar. 31, 2021
Accepted: --
Published Online: Nov. 14, 2022
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20