Application of Different Smoothing Ensemble CARS Algorithm in Spectral Discrimination of Black Tea Grade

Fig. 2. Characteristic variables selected by MA-ECARS based on different window widths
(a): Window width=3; (b): Window width=17; (c): Window width=31

Download full size

View in Article

Fig. 3. Characteristic variables selected by MF-ECARS based on different window widths
(a): Window width=5; (b): Window width=15; (c): Window width=23

Download full size

View in Article

Fig. 4. Characteristic variables selected by GF-ECARS based on different window widths
(a): Window width=5; (b): Window width=19; (c): Window width=31

Download full size

View in Article

Fig. 5. Characteristic wavelengths selected by SPA

Download full size

View in Article

Fig. 6. Characteristic wavelengths selected by CARS

Download full size

View in Article

Fig. 7. The prediction results of GF-ECARS-PLSR

Download full size

View in Article

Table 1. The PLSR model result of different pretreatments

View table

View in Article

Table 1. The PLSR model result of different pretreatments

预处理	训练集		预测集
预处理	RMSEC	$R_{c}^{2}$	RMSEP	$R_{p}^{2}$
Raw	0.246 2	0.970 0	0.410 4	0.916 0
MA Smoothing	0.191 4	0.964 8	0.238 5	0.945 9
GF Smoothing	0.308 3	0.962 3	0.364 9	0.947 7
MF Smoothing	0.233 4	0.947 4	0.275 2	0.927 9
SG Smoothing	0.328 8	0.962 9	0.384 8	0.932 7
De-trending	0.412 8	0.951 9	0.509 5	0.927 8
MSC	0.453 8	0.967 4	0.588 5	0.945 8

Table 2. Characteristic bands selected by moving window partial least squares (MWPLS)

View table

View in Article

Table 2. Characteristic bands selected by moving window partial least squares (MWPLS)

窗口宽度	光谱范围	主因子数	RMSEP
90	862.14~912.07	10	0.288 3
100	861.58~917.04	10	0.276 8
110	859.91~920.91	10	0.270 8
120	854.34~920.91	10	0.264 1
130	839.83~912.07	10	0.255 9
140	833.68~911.53	10	0.246 7
150	833.68~917.04	10	0.237 2
160	833.68~922.57	10	0.230 1
170	828.09~922.57	10	0.227 3
180	816.89~917.04	10	0.220 3
190	811.29~917.04	10	0.214 8
200	810.16~921.46	10	0.210 8
210	796.69~913.73	10	0.204 2

Table 3. The PLSR model of different selection methods of characteristic variables

View table

View in Article

Table 3. The PLSR model of different selection methods of characteristic variables

特征波长选择方法		变量数目	训练集		预测集
特征波长选择方法		变量数目	RMSEC	$R_{c}^{2}$	RMESP	$R_{p}^{2}$
RAW		1 057	0.308 3	0.962 3	0.364 9	0.947 7
SPA		5	0.145 4	0.913 7	0.622 3	0.808 1
CARS		120	0.181 9	0.983 5	0.321 2	0.9 470
MWPLS		211	0.368 1	0.932 3	0.429 7	0.910 2
	MA-ECARS	86	0.246 3	0.969 7	0.265 9	0.965 5
MWS-ECARS	MF-ECARS	142	0.242 3	0.970 6	0.267 7	0.964 4
	GF-ECARS	96	0.232 2	0.973 1	0.251 7	0.969 2

Tools

Get Citation

Copy Citation Text

Li YUAN, Bin SHI, Jian-cheng YU, Tian-yu TANG, Yuan YUAN, Yan-lin TANG. Application of Different Smoothing Ensemble CARS Algorithm in Spectral Discrimination of Black Tea Grade[J]. Spectroscopy and Spectral Analysis, 2020, 40(10): 3254

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Research Articles

Received: Aug. 22, 2019

Accepted: --

Published Online: Jun. 18, 2021

The Author Email:

DOI:10.3964/j.issn.1000-0593(2020)10-3254-06

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. The PLSR model result of different pretreatments

Table 1. The PLSR model result of different pretreatments

Table 2. Characteristic bands selected by moving window partial least squares (MWPLS)

Table 2. Characteristic bands selected by moving window partial least squares (MWPLS)

Table 3. The PLSR model of different selection methods of characteristic variables

Table 3. The PLSR model of different selection methods of characteristic variables

微信扫一扫：分享