A Power Load Forecasting Model Based on CNNLSTM-Attention
Yumeng Zhang, Haoxiang Cao, Zhaofan Zhou · 2024
Under the background of the strategy of “carbon peak and carbon neutrality”, China's green and low-carbon energy transformation is accelerated, and the proportion of new energy generation is gradually increasing, which poses greater challenges to the stable operation of the power system and reliable power supply. In order to ensure the reliable operation of power market, and ensure the electricity demand of industrial, commercial and residential, this paper puts forward the accurate prediction of short-term load based on CNN-LSTM-Attention, the model using CNN initial prediction of time series power load data, and then use LSTM to correct the prediction error of CNN, combined with the attention mechanism, through different parts in the input sequence Different weights are given to highlight the key features and complete the short-term power load forecast.