Research on Short-term Power Load Forecasting Using the WOA-BiLSTM-Attention Model

The foundation of power system operation is short-term power load forecasting, and precise load forecasting can guarantee the power system’s safe and stable operation. This paper chooses the bidirectional long- and short-term memory network model of the attention mechanism under the Whale Optimization Algorithm (WOA-BiLSTM-Attenion) for the evaluation of electric power forecasting in order to address the issues of low precision and poor accuracy of short-term load forecasting. Bi-directional long-short-term memory network modeling attentional mechanisms can lessen the loss of past information and boost the effect of important information. In order to minimize human intervention, the Whale Optimization Algorithm (WOA) is then applied for hyperparameter selection. The experimental results demonstrate that the WOA-BiLSTM-Attention model outperforms BP, LSTM, BiLSTM, and BiLSTM-Attention in terms of short-term power load forecasting, with an RMSE value of 5913.9446 and an R
²
value of 0.9477. Its greater accuracy has some repercussions for other fields' forecasting research.