Fig 1.
The convolutional structure of the TCN model.
Fig 2.
The structure of the N residual module.
Fig 3.
The specific dissemination of the Echo State Network.
Fig 4.
Research framework.
Fig 5.
The learning framework of CEEMDAN-TCN-ESN.
Table 1.
The statistical results of raw data.
Table 2.
Key parameters of the TCN.
Table 3.
Key parameters of the ESN.
Fig 6.
Comparative bar chart of single model prediction evaluation indicators: (a) RMSE indicator; (b) MAE indicator; (c) R2 Indicators.
Table 4.
Evaluation results of single prediction model indicators.
Fig 7.
CEEMDAN decomposition result.
Fig 8.
IMF component reconstruction results.
Fig 9.
Experimental results of the prediction of high-frequency data by each model.
Table 5.
Evaluation results of each prediction model for high-frequency data indicators.
Fig 10.
The prediction process of the TCN model for higher-frequency electric load data.
Fig 11.
Experimental results of the prediction of each model for low frequency plus trend term data.
Table 6.
Evaluation results of each prediction model for low frequency plus trend terms data indicators.
Fig 12.
Comparison bar chart of model prediction evaluation indicators.
Table 7.
Evaluation results of the forecast model on the indicators of electric load data.
Fig 13.
Comparative bar chart of model prediction evaluation indicators: (a) RMSE indicator; (b) MAE indicator; (c) R2 Indicators.
Table 8.
Evaluation results of the forecast model on the indicators of solar power hourly data.