An electricity forecasting method based on multi-type electricity decomposition prediction

With the continuous development of the new power system, accurately predicting the total electricity consumption has become increasingly important. This paper proposes a daily electricity consumption forecasting method based on multi-type electricity consumption decomposition. First, the total electricity consumption is divided into base electricity consumption and temperature-controlled electricity consumption. The base electricity consumption is predicted by fitting it with economic indicators that show high Pearson correlation coefficients. Then, a genetic algorithm is used to optimize eXtreme Gradient Boosting parameters for accurately forecasting the rapidly changing temperature-controlled electricity consumption. Finally, the results of both predictions are linearly combined to obtain the total forecasted electricity consumption. Simulation results shows that the proposed method effectively predicts the daily electricity consumption for a provincial power grid in China, validating its effectiveness.