A Short-Term Prediction Model for Multi-energy Load of IES Based on Comprehensive Meteorological Index and Deep Learning Multi-network Fusion

To improve community load situational awareness, cope with the stochastic and coupled nature of multiple energy load demands, and improve the accuracy of multi-energy load prediction, this paper proposes a short-term prediction model of multi-energy load for an integrated energy system based on comprehensive meteorological indices and deep learning multi-network fusion. Firstly, the basic construction of the integrated energy system is described. Secondly, a comprehensive meteorological index based on biometeorological indicators is introduced and the ReliefF feature selection algorithm is used to construct a correlation model between the meteorological index and loads. Then, a convolutional neural network and Bi-directional long short-term memory load prediction model that introduces the attention mechanism is developed by applying a deep learning multi-network fusion approach. Finally, the scientific and feasibility of the model is verified through the analysis of a case study. The results show that the model proposed in this paper can effectively improve the accuracy of integrated energy system load forecasting.