• Spectroscopy and Spectral Analysis
  • Vol. 42, Issue 3, 749 (2022)
Zhi-xing SUN*, Zhong-gai ZHAO*;, and Fei LIU
Author Affiliations
  • Key Laboratory for Advanced Process Control of Light Industry of the Ministry of Education, Jiangnan University, Wuxi 214122, China
  • show less
    DOI: 10.3964/j.issn.1000-0593(2022)03-0749-08 Cite this Article
    Zhi-xing SUN, Zhong-gai ZHAO, Fei LIU. Near-Infrared Spectral Modeling Based on Stacked Supervised Auto-Encoder[J]. Spectroscopy and Spectral Analysis, 2022, 42(3): 749 Copy Citation Text show less
    The original spectra of corn
    Fig. 1. The original spectra of corn
    The original spectra of yellow rice wine
    Fig. 2. The original spectra of yellow rice wine
    Structural of auto-encoder
    Fig. 3. Structural of auto-encoder
    Structure of supervised auto-encoder structural
    Fig. 4. Structure of supervised auto-encoder structural
    Structure of stack supervised auto-encoder(a): Stack supervised AE; (b): Supervised AE
    Fig. 5. Structure of stack supervised auto-encoder
    (a): Stack supervised AE; (b): Supervised AE
    SSAE training process
    Fig. 6. SSAE training process
    The influence of different learning rates and training times on the model (a), (b): Corn; (c), (d): Yellow wine
    Fig. 7. The influence of different learning rates and training times on the model (a), (b): Corn; (c), (d): Yellow wine
    Prediction results of different modeling methods for corn data(a): PLSR model; (b): BP model; (c): SAE model; (d): SSAE model
    Fig. 8. Prediction results of different modeling methods for corn data
    (a): PLSR model; (b): BP model; (c): SAE model; (d): SSAE model
    Prediction results of different modeling methods for yellow wine data(a): PLSR model; (b): BP model; (c): SAE model; (d): SSAE model
    Fig. 9. Prediction results of different modeling methods for yellow wine data
    (a): PLSR model; (b): BP model; (c): SAE model; (d): SSAE model
    建模
    方法
    预处理
    方法
    玉米数据集黄酒数据集
    RMSEPRPDRMSEPPRD
    PLSR0.235 11.1090.2881.756
    一阶0.182 91.4250.330 11.537
    二阶0.189 81.3730.319 41.588
    SG0.235 11.1090.272 71.86
    MSC1.9990.130 45.7340.08
    SNV0.214 71.2140.789 80.642
    BP0.227 71.1450.172 32.944
    一阶0.130 51.9980.166 993.039
    二阶0.160 51.6240.197 42.57
    SG0.225 81.1550.157 93.212
    MSC0.948 20.274 90.195 12.6
    SNV0.2071.2590.3851.318
    SAE0.197 21.3220.145 93.478
    一阶0.105 32.4770.205 92.464
    二阶0.174 41.4950.194 22.612
    SG0.227 71.1450.172 72.937
    MSC0.181 31.4380.1683.02
    SNV0.231 51.1260.395 61.282
    SSAE0.2641.019 20.1204.227
    一阶0.060 44.3130.367 31.318
    二阶0.189 81.3740.281 91.8
    SG0.263 70.988 80.248 12.044
    MSC0.483 40.539 30.690 40.734 7
    SNV0.313 20.832 30.437 61.159
    Table 1. Prediction results based on different pretreatment methods
    方法玉米数据集
    RMSEPRPD
    PLSR0.182 91.425
    BP0.130 51.998
    SAE0.105 32.477
    SSAE0.060 44.313
    Table 2. Prediction results of corn data sets using different modeling methods
    方法黄酒数据集
    RMSEPRPD
    PLSR0.272 71.860
    BP0.157 93.212
    SAE0.145 93.478
    SSAE0.1204.227
    Table 3. Prediction results of yellow wine data sets using different modeling methods
    Zhi-xing SUN, Zhong-gai ZHAO, Fei LIU. Near-Infrared Spectral Modeling Based on Stacked Supervised Auto-Encoder[J]. Spectroscopy and Spectral Analysis, 2022, 42(3): 749
    Download Citation