Wind energy, a promising alternative to fossil fuels, faces challenges due to its variability and dependence on weather conditions. Effective integration into the power grid necessitates accurate deep learning models ...
详细信息
ISBN:
(纸本)9798350379860;9798350379877
Wind energy, a promising alternative to fossil fuels, faces challenges due to its variability and dependence on weather conditions. Effective integration into the power grid necessitates accurate deep learning models for forecasting wind power generation. While the existing MD-Linear model has shown commendable predictive performance, it struggles with long-term dependencies and complex spatial relationships in wind power data. To overcome these limitations, we propose GMD-Linear, an innovative framework that integrates graph neural networks with wavenet modules. This approach harnesses the strengths of graph convolutional networks and wavenet technology, significantly enhancing predictive accuracy and generalization. Empirical results demonstrate that our model excels at capturing long-term dependencies in wind power data using the advanced Spatial Dynamic Wind Power Forecasting (SDWPF) dataset from Longyuan Power Group Corporation Limited, markedly improving prediction reliability. This method not only addresses the volatility of wind energy but also supports the seamless integration of renewable resources into existing electrical grid infrastructures.
暂无评论