基于麻雀搜索优化的Attention-BiLSTM短期电力负荷预测

Attention-BiLSTM Short-Term Electricity Load Forecasting Based on Sparrow Search Optimization

为了提高电力负荷预测精度,在双向长短期记忆(BiLSTM)神经网络中加入注意力机制。通过对网络的隐含状态赋予不同的权重,减少历史信息的损失,增强重要信息的影响,提高准确性。针对BiLSTM参数选取随机性大且困难的问题,提出了一种利用麻雀搜索算法(SSA)优化的Attention-BiLSTM模型,并通过历史用电负荷数据以及相关影响因素数据进行短期电力负荷预测。首先,对用电负荷数据、气象数据进行预处理。其次,将处理好的数据训练模型,借助SSA对BiLSTM的参数进行寻优,使输入数据与网络结构更好地进行匹配。最后,进行负荷预测。试验结果表明,所构建模型拟合优度达0.9966,有效提高了预测精度且在进行短期负荷预测时具有有效性。

To improve the accuracy of electric load forecasting,an attention mechanism is added to the bi-directional long short-term memory(BiLSTM)neural network.By assigning different weights to the implicit states of the network,the loss of historical information is reduced,the influence of important information is enhanced,and the accuracy is improved.For the large randomness and difficulty in the selection of BiLSTM parameters,an Attention-BiLSTM model optimized by using the sparrow search algorithm(SSA)is proposed,and short-term electricity load forecasting is performed by using historical electricity load data as well as data of related influencing factors.Firstly,the electricity load data and meteorological data are pre-processed.Secondly,the processed data are used to train the model,and the parameters of the BiLSTM are optimized with the help of SSA to better match the input data with the network structure.Finally,load prediction is performed.The experimental results show that the constructed model has a goodness-of-fit of 0.9966,which effectively improves the prediction accuracy and is effective in short-term load forecasting.