基于注意力机制优化长短期记忆网络的短期电力负荷预测

Attention mechanism optimized long short term memory network based short-term load forecasting

在电力系统的经济调度中,如何合理利用电力负荷的过去和现在来推测其未来价值,具有非常长远的社会经济价值.短期电力负荷数据具有明显的时间特征,传统的深度模型越来越多地应用于该领域.然而,深度模型可能存在梯度爆炸或梯度消失,为此,提出了一种注意力机制优化长短期记忆网络的短期负荷预测模型.该模型将长短期记忆网络单元中的激活函数改进为加权激活函数组,并加入注意力机制以提高预测精度.

In the economic dispatch of power system,how to reasonably use the past and present of power load to speculate its future value has very long-term social-economic value.Short-term power load data has obvious temporal characteristics,and the traditional deep model is more and more applied in this field.An improved AM-LSTM short-term load forecasting model is presented.The model improves the activation function in LSTM unit into weighted activation function group,and adds attention mechanism to improve the prediction accuracy.