• Laser & Optoelectronics Progress
  • Vol. 60, Issue 7, 0701001 (2023)
Ying Han1, Kaiqiang Sun1, jianing Yan1, and Changming Dong2、*
Author Affiliations
  • 1School of Automation, Nanjing University of Information Science and Technology, Nanjing 210044, Jiangsu, China
  • 2School of Marine Science, Nanjing University of Information Science and Technology, Nanjing 210044, Jiangsu, China
  • show less
    DOI: 10.3788/LOP213371 Cite this Article Set citation alerts
    Ying Han, Kaiqiang Sun, jianing Yan, Changming Dong. Surface Temperature Prediction of East China Sea Based on Variational Mode Decomposition-Long-Short Term Memory-Broad Learning System Hybrid Model[J]. Laser & Optoelectronics Progress, 2023, 60(7): 0701001 Copy Citation Text show less
    Structure diagram of LSTM
    Fig. 1. Structure diagram of LSTM
    BLS structure diagram
    Fig. 2. BLS structure diagram
    Overall framework of VMD-LSTM-BLS model
    Fig. 3. Overall framework of VMD-LSTM-BLS model
    VMD decomposition diagram of mark point T1
    Fig. 4. VMD decomposition diagram of mark point T1
    IMF0 test set fitting diagram
    Fig. 5. IMF0 test set fitting diagram
    IMF1 test set fitting diagram
    Fig. 6. IMF1 test set fitting diagram
    IMF2 test set fitting diagram
    Fig. 7. IMF2 test set fitting diagram
    Comparison result graph with baseline models. (a) RMSE; (b) MAE
    Fig. 8. Comparison result graph with baseline models. (a) RMSE; (b) MAE
    Comparison diagram of SST fitting effect
    Fig. 9. Comparison diagram of SST fitting effect
    Comparison result graph with ablation models. (a) MAE; (b) RMSE
    Fig. 10. Comparison result graph with ablation models. (a) MAE; (b) RMSE
    ModelParameterValue
    VMDk3
    Alpha1
    Tol10-6
    LSTMLayers3
    Neurous{64,32,16}
    DropoutP0.2
    BLSS0.9
    C10-4
    N1,N2,N3{16,30,280}
    DenseLayers2
    Neurous{10,1}
    Batch_size128
    Epoch300
    Table 1. Algorithm parameters
    ModelMetricsT1T2T3T4T5
    STL-LSTMMAE0.26940.27720.29090.28700.2647
    RMSE0.35180.35880.36780.33450.3240
    EMD-LSTMMAE0.20120.21300.19880.21360.1823
    RMSE0.24910.26120.24390.25640.2349
    MC-LSTMMAE0.25750.26450.24720.25780.2524
    RMSE0.32150.32640.31240.32780.3164
    VMD-LSTM-BLSMAE0.07270.07840.06760.08260.0766
    RMSE0.09570.10890.08640.10320.1166
    Table 2. Accuracy comparison with existing models
    ModelMAE(1,5,7)RMSE(1,5,7)
    LSTM0.25450.33090.39560.28450.43510.4956
    GRU0.25890.35040.40150.31040.45270.5046
    SVR0.31450.49080.68040.41120.60960.8063
    EMD-LSTM0.08650.20120.27350.11870.24910.3045
    STL-LSTM0.13640.26940.29540.18540.35180.3845
    MC-LSTM0.20360.25750.29780.23420.32150.3521
    VMD-LSTM0.06270.11660.17090.08540.14360.1963
    LSTM-BLS0.23660.27850.35870.26880.38470.4423
    VMD-LSTM-BLS0.04570.07270.06690.07420.09570.0885
    Table 3. Comparison of prediction accuracy of models with different time steps
    Ying Han, Kaiqiang Sun, jianing Yan, Changming Dong. Surface Temperature Prediction of East China Sea Based on Variational Mode Decomposition-Long-Short Term Memory-Broad Learning System Hybrid Model[J]. Laser & Optoelectronics Progress, 2023, 60(7): 0701001
    Download Citation