Ultra-short-term forecasting method of demand-side resource response transaction scale based on GAT-Transformer

With the transformation from "source-load interaction" to "source-load interaction," demand-side resources can improve the flexibility of power grid and promote the efficient use of clean energy by participating in power grid regulation. Due to the wide variety and different characteristics of demand side resources, the management of demand side is becoming more and more important. The accuracy of demand side resource transaction scale measurement is one of the keys to improve the resilience of power grid. In order to solve this problem, this study proposes an ultra-short-term forecasting method of demand-side resource response transaction scale based on GAT-Transformer. Firstly, the demand side resource transaction data and meteorological data are input into the GAT network, and the graph feature representation is extracted. Secondly, the feature representation is enhanced by combining category labeling and location embedding, and the dropout operation is applied to reduce the risk of overfitting. Finally, the Transformer module is used to further process the enhanced features, and the Multi-Layer Perceptron ( MLP ) is used to classify and output the final results. The experimental results show that compared with other prediction methods, the prediction results of this method are at least 0.232 lower in RMSE, 0.247 lower in MAE and 5.702 lower in MAPE.