Short-term Wind Speed Prediction Model Based on Secondary Decomposition and SE-SSA-TCN

Background: Aiming to enhance the accuracy of wind speed prediction, a hybrid framework (secondary decomposition-sample entropy-sparrow search algorithm-time convolutional network) was proposed.Methods: This framework integrates SD and SE-SSA to optimize the TCN model.Additionally, a complete ensemble empirical mode decomposition with adaptive noise-variational mode decomposition mechanism was developed to address signal decomposition challenges.A sample entropy strategy was employed to reduce model complexity while maintaining predictive performance.The Sparrow Search Algorithm was utilized to overcome the limitations of traditional optimization methods.Results And Conclusions: Experimental results based on US power station data demonstrated that the proposed model significantly reduced RMSE, MAE, and MAPE compared to single TCN and SD-SE-TCN models, thereby validating its high accuracy and excellent generalization capability.