A Dynamic Combination Model Based Load Forecasting Framework

With the ongoing energy transition, the proportion of distributed energy resources on the demand side continues to increase. Due to its uncertainty and randomness, massive distributed energy resources have brought new challenges to accurate load forecasting. In order to improve the forecasting accuracy, this paper proposes a short-term load forecasting framework based on the dynamic adaptive combination of multiple forecasting models. The framework can decide whether and how to combine various forecasting models for the next period based on the prediction error of each model in the previous period. Based on the load and meteorological data of a region in Jiangsu Province in recent two years, the proposed load forecasting framework is demonstrated to achieve higher prediction accuracy than different single models.