• Spectroscopy and Spectral Analysis
  • Vol. 42, Issue 5, 1426 (2022)
Detached potato leaves after inoculation
Fig. 1. Detached potato leaves after inoculation
The images of leaves cultivated at a temperature of 20 ℃ and 70% humidity for one day to five days
Fig. 2. The images of leaves cultivated at a temperature of 20 ℃ and 70% humidity for one day to five days
Spectroscopic acquisition system1: Reflection probe; 2: Fiber;3: USB4000 spectrometer
Fig. 3. Spectroscopic acquisition system
1: Reflection probe; 2: Fiber;3: USB4000 spectrometer
Detection results of abnormal samples
Fig. 4. Detection results of abnormal samples
Spectral curves of potato leaves with different inoculation times
Fig. 5. Spectral curves of potato leaves with different inoculation times
Changes of POD enzyme activity with inoculation time under the same humidity and different temperature conditions(a): 70%; (b): 80%; (c): 90%
Fig. 6. Changes of POD enzyme activity with inoculation time under the same humidity and different temperature conditions
(a): 70%; (b): 80%; (c): 90%
Changes of POD enzyme activity with inoculation time under the same temperature and different humidity conditions(a): 15 ℃; (b): 20 ℃; (c): 25 ℃ Note: The leaves with inoculation time of 0 days are healthy leaves
Fig. 7. Changes of POD enzyme activity with inoculation time under the same temperature and different humidity conditions
(a): 15 ℃; (b): 20 ℃; (c): 25 ℃ Note: The leaves with inoculation time of 0 days are healthy leaves
The trend of the model root mean square error with the number of variables
Fig. 8. The trend of the model root mean square error with the number of variables
Distribution of characteristic wavelengths
Fig. 9. Distribution of characteristic wavelengths
The results of CARS algorithm
Fig. 10. The results of CARS algorithm
The prediction modeling results of the sample's disease degree established by multiple linear regression analysis (a) and RBFNN (b)
Fig. 11. The prediction modeling results of the sample's disease degree established by multiple linear regression analysis (a) and RBFNN (b)
样本集样本数
/个
范围/
[U·(g·min)-1]
均值±标准差/
[U·(g·min)-1]
建模集9332.94~502.31225.17±117.22
预测集3048.15~477.27237.31±112.95
全部12332.94~502.31228.13±115.85
Table 1. Statistical results of POD enzyme activity of potato leaves
预处理方法建模集预测集参与建模的变量数
Rc2RMSEc/[U·(g·min)-1]Rp2RMSEp/[U·(g·min)-1]
原始0.958 223.831 30.965 039.088 116
中心化0.958 223.831 30.965 023.603 416
归一化0.927 431.409 00.943 641.630 213
S-G卷积平滑0.954 724.819 80.964 739.091 716
移动窗口平均平滑0.950 525.929 60.963 639.140 816
多元散射校正0.980 316.369 10.925 244.951 116
标准正态变量变换0.978 816.994 20.930 141.155 316
基线漂移0.960 823.074 70.967 635.878 316
导数算法(二阶)0.774 755.335 90.434 785.186 02
Table 2. PLS model results of full-wavelength spectrum data preprocessed by different methods
提取方法特征波长/nm
RF547, 697, 707, 740, 754, 755, 758, 759, 767, 814, 878, 879, 888, 941, 942, 946
SPA482, 485, 508, 514, 531, 548, 609, 622, 650, 661, 686, 699, 716, 743, 763, 811, 899, 927, 941, 946, 948
CARS480, 492, 496, 508, 532, 533, 539, 540, 541, 547, 548, 597, 607, 624, 635, 638, 639, 640, 649, 650, 673, 678, 686, 688, 692, 695, 696, 670, 706, 707, 739, 741, 742, 755, 758, 766, 788, 789, 790, 791, 794, 795, 801, 807, 813, 820, 822, 841, 842, 854, 860, 868, 871, 872, 874, 878, 888, 889, 892, 898, 901, 911, 914, 920, 922, 933, 937, 941, 942, 944, 945, 946
Table 3. Feature wavelength extracted by algorithms
模型波长数建模集预测集
Rc2RMSEc/
[U·(g·min)-1]
Rp2RMSEp/
[U·(g·min)-1]
PLS2 4570.958 223.831 30.965 023.603 4
RF-PLS160.736 259.883 70.755 955.417 2
SPA-PLS210.821 049.330 50.942 227.036 0
CARS-PLS720.997 36.032 30.958 125.698 6
Table 4. PLS model results based on characteristic wavelength and full wavelength