Grid Load Forecasting Based on Dual Attention BiGRU

In this paper, the attention mechanism and BIGRU are integrated into the SEQ2SEQ framework, and a new load forecasting model is proposed. In order to make the decoder predictive value more targeted at different time periods, The first attention mechanism makes the model no longer only output a unique intermediate state. The second self-attention mechanism is used to calculate the sequence internal correlation. The validity of the model is verified by simulation experiments based on real coincident data.