1 / 2025-03-03 18:53:09

用于电力负载预测的自注意双向GRU时间序列生成对抗网络

Electricity load forecasting,Self attention,Bi-GRU,TimeGAN

Accurate forecasting of electricity load is crucial for effective energy management and planning. Traditional methods, including LSTM and standard GAN models, often fail to adequately address the inherent nonlinearity and rapid temporal changes in electricity load data, struggling to adapt dynamically to new patterns and anomalies. This study introduces a novel time-series generative adversarial network (TimeGAN) that integrates bidirectional gated recurrent Unit (Bi-GRU) networks with self-attention mechanisms designed for electricity load forecasting. The proposed model leverages bidirectional GRUs to effectively capture the temporal dynamics of power usage by assimilating both past and future contexts. Furthermore, integrating self-attention mechanisms optimizes the network structure, enhancing the model’s focus on significant temporal steps within the time-series data, thereby improving prediction accuracy. Finally, we conducted experiments on a privately constructed power load dataset. The experimental results showed that our model significantly outperformed other network models.