Wind Power Forecasting with a SGMD-K-means-TCN Hybrid Model

The accuracy of short-term wind power forecasting is critical for the efficient management of wind farms. However, existing models often struggle to predict wind power with satisfactory precision. This study introduces a hybrid model leveraging sine geometric mode decomposition (SGMD), Kmeans clustering, and temporal convolutional networks (TCN) to enhance the precision of short-term wind power forecasting. The SGMD algorithm decomposes wind power data into stable intrinsic components, simplifying the forecasting model’s complexity. K-means then clusters these components, with each cluster corresponding to a distinct TCN model that is trained to optimize its parameters. Ultimately, the forecasting from the individual TCN models are aggregated to produce a final forecast. Validated against real-world data, the hybrid model significantly boosts forecasting accuracy, offering a valuable tool for optimizing wind farm operations.